Triangulating molecular evidence to prioritise candidate causal genes at established atopic dermatitis loci

Background: Genome -w ide a s soci a t io n studie s f or a topic derma titi s (A D , eczema ) h ave i denti fied 2 5 reprod uci ble loc i a s socia ted in popul a tio ns o f Europe an de s c en t. W e at temp t to prioriti s e ca ndida te ca us a l gen e s a t t he s e loci u s ing a mul t i fa ceted b ioin fo rmatic app roach and ex ten si ve mole cula r re sour c e s compi l ed into a n ovel pipeli n e: ADGAP P (Atopic Derma titi s G WAS Anno ta t io n & Priori ti sa tion Pip e line ).

W e identi fied a c ompr e h en sive l ist of 1 03 ac ce s s i bl e mole cula r r e sourc es for A D aetio l ogy, includ ing ex pre s si on, pr o tei n a nd DNA me thyla t i o n Q TL data se t s in skin or i mmune- relev an t ti ssue s. Th e se w ere u s ed to te st f o r ov erl ap w ith G W AS s ig nal s (inc luding c oloc alisa tion te st i ng whe re po ssible ) . Thi s wa s c ombined with func t i onal a nn ota tion ba s ed on re gulato ry v ariant predic tio n, and ind epend e nt gen omic fe atur es such a s c hrom atin ac ce s sibili ty, promo t e r- enha nce r inter ac t i on s , s pli cing s it es , non -c oding R NA regi on s, d if f e re nt i al ex pre ssion s tudie s invo lving eczema patie n ts and fi ne -mappin g o f ca usa l va r ia nts. F or each gene a t eac h loc u s , we c on den sed th e e vi d e n ce i n t o a p r i or it is a ti o n s c o r e.

Ac r o s s th e 25 AD loc i inv e stig ate d, w e detec t e d s ig nif ica nt e nrichme n t of gene s w ith ada ptive i mmune r egul at or y func tion an d epi dermal b arri er f ormati on a mong th e top pr i o r iti s ed gen e s . At 8 loci , we were ab l e to pri oriti se a singl e can didat e ge ne ( IL 6R , AD O , PR R5 L , IL 7R , ET S1, INP P5D , MD M1 , TRAF3 ) . A t a fu rt he r 2 lo ci , 2 c andida t e gen e s eme r g e ( IL18 R1 / IL 1 8RAP, LRR C32/ E MSY ). F or t he majori ty of t h e s e , t he prio rit i sed gene h a s b een pr evi ou s l y propos ed a s a pla us i bl e ca ndida te , bu t t h e evide nc e w e c ombine h er e , stre ngthe n s the ca se for many of the s e. In addi t i on , at 6 o f t he 25 loc i, ou r AD GAP P anal ysi s p r i oriti s e s novel alt erna tive ca nd idate s ( SLC 22A5, IL 2RA, MDM1 , DEXI , AD O, STMN3 ), high li ghting t he impor tance o f thi s compr ehe n s iv e app roac h.

Our ADGAP P a n alysi s provi des a dd itio nal suppor t f or previou sly imp lic ated ge n es a t seve ral AD GW AS loc i, a s w ell a s ev idenc e for pl au sible novel ca ndi dat e s at o ther s. W e highli ght seve ral gene s wi t h good /conv e rging e videnc e of i nvolv eme nt in A D tha t rep re se n t pote ntial n ew targe t s fo r dr ug disc ov er y .

ato pi c derma titi s, ec zem a, im mune dis ea se, GWAS , col ocal i s a t io n, gen e prior i t i sati on, fine -ma pping, e Q TL

De fined by infl a med dry, hype rpla stic e cz ematou s s k in and pru r itu s, atopic de rma titi s ( AD) i s a mong the w or l d’ s top 50 co mmon dis ea se, w it h preval ence in 2010 e stima te d at c lo s e to 230 mil lion c ase s and increa sing [1 ]. AD i s p a r tic ul arly common in ch ildhood, w ith 10-16 . 5% prev ale nc e in the first 5 ye ars of l i fe [2,3 ], 80% of wh ich w ill then pr o gre ss in th e so-c all ed at opic march a n d deve lop oth er all ergic conditio ns [4 ]. In 2 015, the bu r de n of A D in term s o f co st bo rn by i ndivid uals, thei r famili es a n d s o cie t y at l ar g e , was es t i m a t ed a t cl os e t o 5 .3 b i ll i o n d o lla rs [ 5 , 6] . A D co m o r b id i t i es i n cl u d e . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. co ndition s s uc h a s: all ergic r hi niti s, a s th ma, allergi c c onjunc t iv iti s , e os inophilic e s ophag iti s a nd food all ergy[7 ]. Due to the s y s t emic immune natur e of the di se a se, AD i s al s o a ssocia te d with i ncrea sed risk f or many infl amma t o r y di se as e s : rh e um atoid ar thri ti s (R A), i n f l amma tory bo w el di s e a se ( IBD ), sys t e mic l upu s e r y t h ema to s u s (SL E) an d decrea sed r i sk f o r typ e 1 dia be te s (T1D) [ 8,9]. AD i s highly herit able - with e stima te s o f up to 7 5% in t w in stu die s [10] . The large s t and mos t rec ent gen ome-wid e a ssoc iati on s tudy ( GWAS ) of A D, unde rtak en by the EA G L E c ons o rti um in 2015 iden t i fi ed 25 l oci ( ex plai ning 14 .9% of th e varianc e ) tha t we r e a s s oc ia ted w it h AD in indiv idual s o f Europea n d e scent a nd replic a ted in a n indep end ent sample [11 ]. The ma jori t y of d ise as e a s socia ted va r i ant s a re loca te d in non -c oding r eg ion s. Thi s ob se rvation impli es tha t many dis ea se -a s s oc ia ted va r i ant s ha ve a re gulat ory role r ath er t h a n aff ec t i ng pr otein func ti on. I t ha s now b een widely recog ni sed th at in tegra ting variou s bio lo gic al data r e s o urce s c an prov id e c omple mentary evide nc e abou t GWAS c au sa l ge ne s[12 ]. Acro ss se veral dis ea se i t ha s bee n e s tima t e d that ~50% of th e time, the g ene through w hich t h e ca u s al SN P a c t s i s no t the c lo s est g ene to th e i ndex S N P [1 3,14 ]. The EAG L E GWAS u tili sed small e QTL re s ource s (Tw in s UK L CL a nd skin mic roa rr a y sample s [15 ]), fine - mapp ing tool s (M AN T RA[16 ]) , varian t a n nota tion softw ar e (Regul omeD B[17 ] and Hapl oReg [18] ) an d two d if f e re nt i al ex pr e ssion stu die s on ec z em a to inv e s tig a te po ten tial cau sal g e n es a t th e ide n t i fi ed loc i. Si nce 2 015 ther e h a s be en an expl o s ion o f new da ta s e ts from many ce ll type s ( such a s e QTLGen [19 ], full GTEx r e lea se [20 ], Bl ue print [ 2 1] , Go D M C, pr omo t e r- enha nce r H iC da ta set s ) and new method s ( s uc h a s finema p [22 ], J AM [ 2 3] , fa s t P a into r [2 4 ], TW AS[25 ] ) bein g made ava ilabl e t h a t of fer an o pportuni t y to re fine pri oriti s ati on of ge ne s a t th e GWAS loci . Se veral e xce lle nt c omput ation al a ppr oac he s to a id ide nti fic ati on of cau sa l g ene s a t GWAS loci exi st, but the s e meth od s we r e no t deem ed s ui ta ble for ou r purpo se a s they e i t h e r appl y a single meth od ac r o ss a la r g e numbe r of GWA S ( such a s S MR[26 ] and P henom eXc an[27 ]) bu t do n ot int egrat e ev idenc e f r o m acro s s many so urce s, a r e out of d at e (e .g . G W AS ra p[28 ], la s t upda ted i n 2013), as s e mbl e a lot o f da ta b ut d o not show c l ear t arget p riori ti sati on s uch a s F U MA [2 9], I N FE RNO[30 ], o r their sou rce da t a s e ts are le ss r eleva n t to our dis ea se in term s o f ti ssue sel ec tio n bec au se th ey t a r g e t anoth er spec i fic d isea s e, e.g. I NQUIS IT a nd brea st canc er [ 3 1] . In thi s pap e r , we a im to c ompre hen sively dis s ec t AD G WAS l oci by prio riti s i ng ca n didate c au s al gen e s a nd ill umina ting biologi cal mecha n is m s thr ough w hich cand idate ge ne s can impac t AD ri sk . We app lie d a number o f fi ne -ma pping a p proache s a nd in tegr ate e videnc e from ac ross many s ou r c es relev an t fo r s k in di se a se. We de vel ope d AD G AP P ( Atopic Derm atiti s GWA S Anno t ation & Priori ti sa tion Pip e line ) wh ich of fer s a c on s i s t ent me thod o f prio r i t i sing A D candi da te ge ne s an d va r i ant s by dev is in g a r ank ing sy st em. A D G AP P whic h ex plici t l y model s our a ssum ption s abou t th e importanc e of di ff er ent typ e s o f e vide nc e a s w ell a s st r e ng th of the a s socia tio n s r ela t i ng th e f ea ture s to ge ne s and va r i ant s. In ou r G WAS t arge t priori ti sati on pip elin e, we us e me t h od s (c oloc, TW A S ) t o fo rma lly compare t he as s o cia tion pa tte r n s in Q TL studi e s and G W AS to te s t for their c ol oca li sati on whe n ever full s umm ary sta t i s t ic are av ailab l e, a s ~50 % o f c ommon va r i ant s a re a s s oc iat ed w ith one e QTL or more in GTEx [ 3 2] s o s imp le look up s for v aria nt o ve r l ap alo n e wil l resul t in many fal s e po si tives . We a l so integ r a te the re s u lt s o f two indep ende n t pipeline s : e nr i chmen t-b a sed r egfm [33 ] and netw or k -ba se d PrixF ixe [ 3 4 ] fo r G W AS -wi de prio riti s a t i on of ta rget ge n e s . We ai med t o s elec t r el evan t da t a s e ts to r eflec t th e full mole cul a r a e tiology o f AD . Broadly s pea king, AD pa thoge n esi s i s t hought t o s tem from epid ermal ba rrie r di sfunc tion, a s well a s abnormal immune sys t e m ac tiv ation [35 ]. In a ddition to skin lesi on s , A D abnormal itie s a re al so s e e n i n non-l e s i o nal skin and blood [36,37 ] . Ther efo re, w e mad e it a priority t o includ e molec ula r QTLs f ro m skin, bl ood and . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint immune - s pec i fic cell type s in A DG A P P. D irect imp ortanc e in AD p hy s i ology is eval uated by inclu ding diff er ential gen e expre s s io n and D NA me thyla t i on studi e s a nd p rot eome c ompari s ons involv ing ec z ema pa tient s . As t h e le ad SN Ps id enti fie d i n the ini tia l GW AS a naly s i s ma y not b e c au sal [38] , we us e thr ee Ba ye si an fine -ma pping a lgori thm s t o ini tiall y priori tise ca u s al SN P s .~ 90% of index SN P s a r e plac ed in the n on- co ding regi on s of t h e genom e , and so c a usal v ari ant s are expe ct ed to ha ve ma i nl y reg ulatory func t i o ns[18 ]. I n orde r t o incl ud e the fun c t i onal po ten tia l o f ea ch v arian t in ou r r a nk ing, we direc tly look up ov er l ap o f pot enti a l GWAS cau sa l variant s with e xperim en t a lly g en era ted func t i o nal anno t a t io n s a s w ell a s pr edi ction s for re g ulatory i mpact o f th e varian t g ener at ed by mac hine lea r ni ng mode l s. I nte gra tion wi t h high -th roughput c hrom atin c on for ma t i on ca p t u re ( 3 C) da ta al low us t o che ck if our varia n ts o f inte re st are loc ated in e nhanc e r region s or p romot er s o f g ene s hig hlighted i n e QTL a naly si s in the ma tch ing tissue . We int eg r a te sev eral e stabli sh ed fi n e -ma pping and g ene prio riti sa tion me thod s in a unique AD- focu sed gen e priori ti sa tion pi peli ne t o co mprehen sive ly e valua t e th e ca u s al gen et ic ev idence at e ac h loc us and u tili se a n exha ustiv e set o f A D- rela te d molec ula r dat a set s t o be st sup po rt th es e meth od s. Our pipelin e in c ombining the se metho d s gene rat e s a g lobal s c or e , whic h c an be u s e d t o a sse s s th e mag nitude o f ev idenc e for (and r ela t i ve ev idenc e be twe en ) te sted gen e s a s the c aus al ge ne a t a partic ula r loc u s . S uch a s c ore c a n s er v e a s a met r ic wh ich a llows ra pid gene p r i orit isa tion by mole cular bi ologi s ts a nd oth er int er es t e d partie s , s uc h a s ph arma ceu tic al c ompa nies .

Source GWAS Pat erno s ter et al . (20 15) [11 ] is the big ge st G WA S on A D to d a te, co n s i sti ng of E AGL E Con sor t i um data o f 21,399 ca se s and 9 5,464 c ont r o l s from pop ula tion s o f mo st l y Europe a n, b ut al so A frica n, Japan e s e and L ati no anc e s t ry. In ou r ana l ysis, we i nves tigat e 25 loci , whic h show e ither gen ome - wi de signific anc e and for nov el lo ci a re rep lica ted in inde p enden t Europe an anc e stry s amp l e (21 loc i), or a re signi f i can t lo ci prio ritiz ed by t h e MA G E N TA g en e set enric hmen t ana l ysis[39 ] p res ent ed in the or i ginal p aper (4 l oci) [11 ]. Bayesian fine-mapping To i denti fy li kely cau s al gen etic va r ia n ts i n t he regio ns h arbou ring AD GWAS signa ls , u s in g only informa tion on a s s oc i ation s tr e ng th and L D s t ructu re, we u sed thr ee di ff e r en t Ba yesian fi ne - mapp ing method s: F in emap [22] , fa stP ai nt or [24 ] a nd JA M[ 2 3 ] . Eac h method r eli e s o n di ff eren t prio r as s u mp t i on s a nd model for m ul ation . As inpu t da ta we us ed, the a ssoci a tion sta t i s tic s f rom the AD GWAS in i ndivid ual s of Europ ean anc e st ry, publi sh ed in Pa te rno s t er e t a l. ( 2015). For LD s tr uc tu r e , in F in emap a nd Pai ntor a naly si s, we us e d r cor rela tion s c alcu la ted from th e 100 0 Genome s [40 ] Europe a n re fe r e nc e ( n = 503), a s i t all owed i nclu sion o f mor e high con fiden c e S N P call s c ompa red t o UK Biob ank [41] - 9 ,265,840 ver su s 8,39 1,826. UK Bi obank pan el[41 ] fil t e r e d for Euro pean a nce str y ( n =48,1 67 ) with st anda r d Q C appl ied [42] w a s u s ed a s LD re fer ence pa nel in JAM a nal ys i s. T his wa s bec a us e JA M r e quir e s a high numbe r o f individ ual s in the r ef er enc e a nd r e moval of highl y co r r ela te d SN P s f o r the geno type matrix t o be inv er t ib le . To t ha t end, w e a lso pru ned SNP s prio r to feed ing them to JAM. W e u sed a thre s ho ld of r 2 > 0 .95 in Prio rity P rune r ( http: / / prio rityprun er .s ource forg e.n e t/ ) a nd s e t mi nor all ele . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint freq uenc y (MA F) th re shol d at 0.05 and m inimum S N P call r a te of 0 . 9, while f or c e s elec t i ng al l the inde x SNP s. Our fin e -mappin g int egrati on pr o toc ol in v olved r un ning a ll the 3 pr og ra ms wit h al l t h e SN P s within the in terv al o f 10k bp, 100 kbp, 5 00kbp, 1Mbp, 3Mbp c en tred on the i nd ex S N P . We als o us ed r 2 - bas ed a nd D ’- ba sed hapl oblock inte rval s defi n ed wi t h t h e BigL D [43 ] and Gpar t [44 ] alg or i thms , re spectiv ely, i n 1000 Gen ome EUR pa ne l. We ran ou r F inemap an aly s i s u sing shotg un s t och a s tic s e a rch, w her ea s fo r Pai nto r, we varie d th e alg orit h m f r om exa ct ex hau stiv e s earch to MCM C when c on sideri ng from up to 2 or max . 5 c ausal SNP s in th e reg ion , ac cording ly. Ma ximu m n umber of alg o r i t hm it era tion s wa s se t at 1000 in Pain t or and 10 mill ion in JAM . In or d e r t o p ermit analy s i s o f a bina ry t rai t i n JAM, linea r ma pping o f log -odd s ratio s wa s per fo rmed [45] a nd the re sidu al v arianc e inv erse g amma h ype rprior s w ere se t t o "G a u ssia nRe sidu a lV arianc e InvGamma P r i o r _a " = 2, " G a u ssian Re sidua l Va rianc e InvGamma P r i or_b " = propor tion o f c a s e s * (1 -p r op ortion o f ca se s ). When c omparing ou tput of Fin em ap, P ai nt o r a nd JAM we o nly c on s i de red t op fin e-ma pp ed SN P s wi t h Baye s Fa cto r > 100 and po st erio r pr obabili t y o f being ca u s a l o f at le a s t 0.1 . Variant filtering In sub s equen t ge ne a na lys e s , s how n bel ow , we limited ourselv e s t o SN P s w ithin the reg io n in signific ant L D w ith the ind e x S N P i n 1000 G en ome s EU R pop ulati on, henc e fo r th r e fer red t o a s the GWA S loc u s i n t e r v al . Th e region in eac h ca s e wa s de fin e d by the furth e st re move d S N P with r 2 > = 0.2 wi t hi n 1 Mbp in t e r v al c entre d on th e in d ex S N P and al l t he SN P s within t h e bound ary were co ns id e r e d . We con f i r m e d tha t we captu red a ll th e S NP s wi t hi n the b roadly de fin ed haplobl ock by re- de fining the bo unda r i e s b a s e d on max imum 3 M b p interval . D e finiti on of ou r h a plobloc k c hanged only in the c a se o f 3 i ndex SNP s : h o weve r i n all cas es the re wa s a str etch o f a t le a st 100 k bp (r s 61 813875 with a very sha rp L D decay and rs41293 864 sit uat ed in t he MH C reg ion w it h compl ex LD), u p to 350 k bp (r s 777 14197) o f S N P s w ith no r 2 > 0.2 so we di s mi s s e d tho s e a s outlier s an d us ed the 1 M b p int erval -de f i ned region in al l c a ses . Identification of key tissues and cell types In orde r to foc u s on key ti ssue s / c ell ty pe s a ssoci a ted wi t h ecz ema v arian t s, we u sed gene se t enric hmen t in S N PS ea [46 ] w ith the suppl ied gene expre s s i on dat as et s: Gen e At l a s Affyme t rix ex pr e s si on in 79 huma n ti ssue s [47 ], Imm unol ogica l Gen ome P r oje ct [48 ] Af fymet ri x e xpres s ion in 249 mur i ne blood cel l type s a nd F AN T OM CA GE[49 ] in 533 huma n cel l type s. We ran S N P S ea u sing recom mended s et tings a nd u sed index S NP s a s i nput. SN P s e a co n s id e rs gen e s i n LD w it h the ind ex SNP s and for e a ch l ocu s and c ell -type co mbi nation, sel ec t s on e gene whic h sh ows highe s t ti s s ue - speci fi city o f expre s s ion . The me r g ed ge ne s et acr os s l oci i s th en s c or e d for i t s c e ll-type s p ec ific ity and t he r e s ul t in g scor e c ompared to th e one s ob t a ined from th e nul l di str ib u t i on of re s u l ts fo r random S NP set s ma tch ed on th e numbe r of gen e s i n LD in o r de r t o ob t a in a pe rmut a ti on p -val ue. Se condly , we use d MAGMA [50 ] ge ne enr ic hment ana ly si s o n G T Ex 7 .0[20 ] dat a a s c arried ou t by FUMA [29] . Brie fly , MAGM A ge ne -ba se d ana lysi s wa s done u s ing SN P s in th e loc u s int erval to id enti fy all the gen es a ss ocia te d with o ur GWAS hi t s , and s ub sequ e nt l y MA G M A g ene -pr operty an a lysi s wa s appl ied t o te s t t i ssue typ e s p e cific i ty in e xpres s ion o f ide nti fied gen e s. TwinsUK eQTL identification . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint We u sed ge no type a rray dat a a nd RPKM -normali s ed e xpre s s ion in ly mphobl as toi d c ell li ne (LCL ) a nd skin ti s s u e for femal e s in t he Twin s UK co hor t [51 ]. RPKM val u e s w ere r ank- tran s fo r me d to no rmality using G enABEL [52] R pa cka ge b e fore e QTL mappi ng. ci s - e Q TL s 1 .5 Mbp up str ea m and downst r e am of TS S w er e id enti fied u s i ng line ar mixe d model implemen ted in GEM MA [ 5 3] . We use d ag e a s co variate i n the an a lysi s involv ing a ll sam ples and cent red r ela te dne ss ma trix as ra ndom ef fec ts . PEE R anal ys i s w a s r un t o iden tify any a dd itional hidd en c ovaria t e s not c ap tur ed ab ov e[54 ]. e QTL as s o cia tion s w ere iden tifi ed u s in g the W ald t e s t . CEDAR eQTL re-identification In the an aly s i s inv olving the CEDA R c oho rt ( e .g. Momo zawa et al. (2018 )[55 ] ), we us e d the publi cly av ailab le da ta: impu ted ge no t y pe s and n orma liz e d gen e expre s s i on val u es fr om b lood and int e s tina l ce ll type s ( CD4 + T lymphoc yte s, CD8 + T ly mphocy t e s, C D 19 + B ly mphocy t e s , CD14 + m onocy t e s, C D15 + granul ocyte s, pl a tel et s , il eum, c olon, rec t um) adjus ted for 4 to p P C s a nd cov aria te s ( sex , ag e, smoking sta tu s, batc h ). We u s e d GEMMA’s li n ear mix ed model a nd Wald t e s t to re-i den tify cis -e QTLs wi t hi n 1.5 Mbp up s t ream a nd dow n stre a m of T SS. Colocalisation with coloc and TWAS We obt ain ed full summary stati s tic re sul t s f or c is- e QTL s de tec ted i n whole blo od i n the e QTLG en data se t [56] – a cc e s s ed on 08/08 / 2 018, e QTL s from GTEx ver. 7 data se t iden tifi ed in the f ollo wing tis s u e s: w hole blo o d, s p le en, sun -exp o s e d and une xpos ed skin, tr an sf ormed fibro blas t s a nd EBV - tran sforme d lymphoc yte s[57 ], e QTL s pu bli s h ed from t he Kim-H ellmuth e t a l . (20 17 ) study inv es t ig ating mon ocyte re spon s e to mic ro be- a s s oc iat ed mole cula r p at t e rns [58 ], e QTL s in t h e monoc yte s, neu trophi ls a nd C D 4+ T c ell s from th e BLUEPR I N T pr oj ec t[21 ], a n d p QT Ls fr o m whole blood in the S un et a l. (2018) d ata s et [59] as w ell a s Tw insU K and CED AR e QTL s id e nt i fi ed abo ve (Da ta se t S1) . S ub s eq u ently, c oloc ali s atio n signa l be twee n b eta s fr o m GWAS an d e QTLs/ p Q TL s for gen e s w ithin 1.5Mbp up s t re am and dow ns t ream o f index SN P wa s e valua te d wit h the coloc [60] R pac kag e. In a ddi tion , thr ee -wa y c oloca lis ation o f GW AS and whole bl ood mole cul a r p heno type s: pQTL s and e Q T L s was inv e stig ate d wi t h moloc[6 1 ] with de fa ult p rior s bu t broug h t no signi fic an t re sult s. In col oc an aly si s, we con sider ed loc i w ith posteri or prob ab ility o f H4 (P P H4) > 0.5 a s informa tiv e eno ug h to be in clud ed in AD G AP P , a s don e pr ev iously [62] ; w ith H 4 st ating the hyp ot he s i s o f both trai ts b eing a ssoc iat e d and s ha r i ng a si ng le ca u s a l va r ia n t . We al s o c ar r i ed ou t a TWA S [25 ] ana ly sis, w here re fe renc e da ta s et s with ge ne ex pr e ss ion and gen ot y pe da ta ( G T Ex v er. 7.0, CE D AR a n d Tw ins UK) w er e u se d to pr edic t g en e e xpre s s ion in our targe t GW AS. W e u sed 10 0 permu ta tion s to con s ervativ ely cali bra te t h e impu ted g ene e xpre ssion as s o cia tion s t ati s t i c c ondition al on th e G WAS s t r eng t h o f a ssoc iati on a nd u s e d p r ovided Europea n1000 Genom e s p a nel for L D ref e r en ce . Any s i gnific an t gene expr e s s i on a s s oc ia tion s with AD wer e the n p ost -proc e ss ed to id e nt i f y con ditionally ind epe nd ent a s s oc ia tion s . I n a ddition, c ol oc ana lysi s wa s ca r ri ed out , ba sed on ma rginal TWA S wei ght s with provid e d sc ript s . The analy s i s pip eline for t he S MR an al ysi s ha s b e en de scrib e d previou sly [26 ]. In b rief , e QTL data s et s from th e e QTL Gen cons ort i um[19 ], BLUE PRI NT[21 ] ac r o s s 3 ce ll type s (monoc yte s , neutr ophil s and T ce lls ) and GTEx ver. 7 [ 2 0 ] ac ros s 48 ti s s u e type s were downlo a ded a nd. The S MR method app li e s th e Wald Ra ti o meth od sy st ematica lly for a ll gen e s w ith an e Q TL at P< 1x10 -4 , using l ea d eQTL a s inst r um en tal va r i abl e s and ecz ema e stim ate s a s th e outc om e. The HEI DI (he ter ogenei t y in depe nd ent in strume n ts) t e st w a s appli e d to filt er ou t ge ne tical ly predic ted e f fec ts wh ich ma y be att ribut ed to h ete roge n eity in a region w hic h may lead t o spuriou s r e sult s. Regfm and PrixFixe analysis . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint To furth er pri oriti z e G W AS gen e ta r g e ts, w e used two g ene pri oriti sa tion me thod s : regfm [33] and PrixF ixe [ 3 4 ]. PrixF ixe st rat egy rel ie s on p r ioriti s a t io n o f group s o f ca ndida te g ene s fr o m multipl e GWA S loc i ba s ed on ' cofunc ti on' netw ork s ( CF Ns ). A gen etic a lgori thm optimi sa ti on a ppr o ach sel e ct s optimal s e ts c on s i s ting of on e gen e p er l oc us w her e th ere ex i sts a d en s e s ubn e t w ork of f unc t i on al relati on ship s wi t h gene s a t oth er GWAS l oc i. Ea ch ge ne i s then sco red t o re fle ct i t s cont ributi on to the to p sub netwo r k s. Re g fm’ s workflow i nvol ves int er s e c tion o f fin e - m ap ped c re d ible i nt e r v al S N P s wi t h cons en su s DHS sit e s and g ene s who se ex pre ssion th ey control p redi ct ed ba s ed o n ROADMA P[63 ] c hrom atin ac ce s sibili ty an d ge ne ex pr e ssion da ta to pr i or i t i se t arge t gene s. Variant functional prediction KGGSeq [64 ] w as u sed to me as ure no n-c o ding v ariant r egula to r y poten tial a n d co ding v ariant dele te r i ou s n e s s us i ng func t i o nal score s d erive d by c ombining scor e s from CADD [6 5], D AN N [66 ], Fun seq2[6 7 ], fat hmm-MKL[68 ], GWAS3 D [69] , S uRFR [70] , GW AV A[71 ] algo rithm s. The new e xtende d ve r sio n of fa thmm : fa thmm-XF [72] , GW A S4D[73 ] and fitC on s[74] we re al so u s ed i ndepend en t l y. Overlap with ChI P-S eq de fine d binding s i t e s of tran scri ption al regula tor s wa s c r o s s-re fe renc e d in t he ReMap2 018 da ta ba se [75 ]. Spl i cing r e gul atory po t e ntia l of v ari ant s wa s eva lua ted w ith SPI DEX[76 ]. We al s o l ooke d a t varian t ov erl ap w ithin diff er ent r egul ato r y region s: in sul ator [77 ], promo te r- enha nce r inte rac tio ns[7 8 ] , [79 ] , [80 ] , [81] , [ 82] , [83 ] , [84 ] , [85 ] , [86 ] , reg ulat ory non-c o ding RNA s[87] , [88] , [89 ] , [ 90] , [91 ], to polo gica lly a ssocia t i ng doma in s (T A D s ) [ 9 2 ] , [93 ] , [79 ] , [94] , [ 81] , [95 ], a nd CTCF bindin g s i te s [96 ] c ulled f rom variou s public a tion s, u sing g iggle [97 ] se arch en g ine. Indep enden tly, we l ooked for ov erl ap in s ide Roadmap reg ion s cla ssifi ed a s c ontai ning ac t iv e ch r oma t i n s t ate (sta te s 1 -8) [63 ] and FA IR E-Seq -de t e r mi n ed regio ns o f a cce ssibl e c hromatin i n huma n epide rmis du ring barr i er matu rati on and disrupti on [98] . Cell - type spec if ic r e gulato ry ele men t s [99 ] w er e al so a nno tat ed ba se d on his t one mark s and c hr omatin sta te . Independent lookups We hav e al s o pe r f ormed gen e a nd va r ia n t lookup s am ong publi s hed s ig ni fica nt re s ult s from vario us e QTL[21 ,51,99– 125 ], m QTL ( i ncludi ng G o DM C r e sult s [126 ] ) [121,124 ] p Q TL [127,12 8] , , h QTL[2 1,129 ], ca Q T L[130 ], w he re f ull G W AS re sul t s wer e not av aila bl e, a s w ell a s di ff eren tia l ex p r e ss ion [131–1 35 ] , DN A me thylati on[13 6 ,137] a n d prote om e[138,13 9 ] compa r i s o ns b etwe en skin in AD pa tien ts and hea lthy c ontrol s ( Da ta se t S1). We a l s o in terrog at ed the GW AS Ca talo g [140] (ac ce s s e d on 11/01/2019) for a ny va r i ant s th at h ave b e en ide n t i fi ed a s gen ome -wide s ig ni fica n t in previo u s G WAS st u d ie s on r elat ed i nfl ammato ry c onditio ns. We u sed the sign ifica n ce th re shold d efin ed in e ach pape r , as i t varied du e to a numb er o f com parison s ma de. Generation of candidate gene and SNP rankings The r e s ult s o f ana ly se s and l ook up s lis t e d above were th en int egra ted t o prov ide two ran kings o f: 1 ) all the SN P s within e ach G W AS l ocu s inte rval and 2) all th e gene s withi n 3 Mbp w i ndow c entre d around ind ex SN P . Thi s wa s a chie ved by giv en a sco r e t o each pi ec e o f ev idenc e a nd summing ac r o s s the se sou r c e s to gen e r a te a ca u s a l p riori tisa t i on s c o re f or eve r y SN P and eve ry g ene te s ted. Th e se s c o res r e p r e s e nt t he s t r en g t h of ev i d e n c e f o r a ca usa l r o l e of t he S NP o r g ene i n A D b as ed on t h e ev idenc e a s s imi lat ed. . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint Det aile d me thod o f c alc ula tion o f ba s ic s core pe r g en e or v aria nt in a giv en ex peri ment/a naly si s is pre sen ted in Add i t i on al F ile 1: Suppl em e ntary Tex t. Br ie fly, t he s c or e is p ropo rt io n al to th e mag nitude o f re sul t e ff ect or signi f i canc e , dow n-wei ght ed ba sed o n t o ta l SN P/ge n e hi ts in a loc u s i n the ex perim ent /ana ly s i s, n experime n ts in the s tudy , and up - w eigh ted ba se d on p re- a ssign ed ev idenc e wei ght o f th e s tudy ( Dat as et S2 ). F inally , a ft er summing u p s c or e s from i ndividu al ana lys e s / ex perimen ts, t h e s c o re i s fu rt h e r adju st e d b a sed on a ve rage numbe r of stu die s and stu dy type s sh owing evi dence fo r a giv en SNP / gene . We dow n-weig h t e d evi dence f r om any si ngl e ex perimen t/ a nalysi s whi ch hig hlight ed multiple gen e s /va r i a nts a s bei ng les s s p e cific . Do wn-we ighting ba s e d o n n e xperime n ts in the s tudy wa s introduc e d s i nc e e xperime n ts in th e same study are no t in dep enden t, a nd so we w anted t o att enua te t h e import ance o f multipl e hit s to e xperimen t s within th e sa me s tudy. We al s o sca led the scor e by the w eight o f e videnc e , ie . subjectiv e p rior b elie f in s treng t h o f th e ev idenc e. W e st r ongl y prioriti s ed hit s wit h ev idenc e we ight 1 ( formal t e s t of a s soc iatio n u s i ng th e full set o f summary sta ti st i c s f r om expe r i me ntal da ta - her e fin ema pping, c oloc , T WAS) . The oth er two e videnc e w eight c a tegori e s inc luded di r e ct look up s among signi fica nt r es ult s obt a in ed i n ex periment s a nd predic tion s u s in g a ma chine lear ning model t r ain ed on ex perime ntal da ta, f or inst ance varia n t p a thogenic i ty predi ctio n . La s tly , the t otal scor e fo r a give n gene or SNP wa s a dju s te d by the a ver age numbe r of studi e s and st u d y type s sh owing evi dence f or a giv en SNP/g ene , whic h aim ed to promo te h et eroge n eity of ev idenc e s o ur c e s a nd thei r type s , suc h a s dif f eren t type s o f mole cula r ma r k er s. The de ci sion s made in c alc ulating t h e final s c ore r e flec ted var ious tr a d e- of f s . F o r exam ple, for ge n e ra nking s , we w ant ed to rewa rd consi sten t a ssocia ti on o f a g iven gene with SN P s in L D with le a d va r i an t, but a t th e s ame time di d not wa n t t he s c o re to b e a rtific i ally infla ted by repe a te d a s s oc iati on s of diff er ent v arian t s t o t h e s a m e g e ne in j u s t on e s t u d y.

Identification of key tissues and cell types in AD GWAS loci Sk in, as t he prima r y af fec ted organ , a nd blood cell t y pe s, giv en th e e st abli shed i m mune c omponent , are o f rel evanc e fo r A D[141 ]. We al s o a t t empted t o iden tify addi tional t i ssue s o f importanc e by using g ene - se t enrichm e nt me thod s on g ene expre s s i o n data s et s a cr oss a vail a ble tis s u e / c ell type s t o iden t i fy any t i ssu e/c ell -sp ec ifici ty enrich e d in ge ne s linke d t o our G W AS loci . MAGMA g ene -pr oper ty ana ly sis o n 53 ti s s u e type s f rom GTEx ver. 7 id enti fied sig nifi cant enric hmen t (at p < 0.001 ) fo r ge ne s with ti s s u e- sp ec ifi c e xpres s ion in EB V -tr an s f ormed lympho c ytes , whol e blood , s pl e en, sun -exp o s ed and n on- ex pos ed skin a t o ur GWAS lo ci . D es pite n ot r eac hing stati s t i cal signific anc e, we al so inc lud ed tr an sf or m e d fibr obla st s ( p = 0.1 ), du e to t he rol e of de r ma l fib robla s ts in ski n maint ena nce a nd rep air [142] . SNPS e a enric hm ent an a lysi s i n th e G ene Atla s da ta se t priori ti sed (at p ermut ed p <0.05, 79 ti s s ue s te ste d) whol e blood a nd blo od c ell t y pe s : CD4 + T c ells , C D14+ monocy t e s, C D8+ T ce lls, an d de ndri t i c ce lls . In add i t i on, Immun ol ogic al Ge nomi c s dat a s e t ( 2 49 ce ll type s) priori ti sed Nat ur a l Kill er T ce ll s, whi le FANTOM CA G E data s et (533 c ell ty pes) p r io riti s ed s timulat ed monocy te s, b a s op hi l s , M a s t c ells , eo sinophi l s , n eutroph il s , Lang erh an s cel l s , C D 34 + p rogeni tor c ell s, h air f oll icle ou t er r oo t s hea th cell s as we ll a s sple en. Fo r tha t re as on, we de ci ded to inc lude d a ta set s inv olving all possible t y pe s of . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint immune cel l data s et s in our an a lysi s , in a ddi t io n t o s k in ( inc luding fibro bla st), sple e n and whole blood . We revi ewed t he li ter atur e to id enti fy 103 sepa rat e d a ta s e t s from the se ti s s ue -ty pes w ith r e leva n t data f o r th e pr i o r i t i sa tion o f gene s (and S N P s) . We i d enti fie d 83 SNP -ba se d dat as e ts w ith ind i vidual va r i ant -ba s e d (molec ul ar QTLs, va r i a nt fu nctiona l pr edic t io n s o ftwa r e ) or int erval - bas ed (T AD, promot er -enha nc er int erac tion s ) info rmation t h a t we u sed to a id va r i an t prio rit i sation . We iden t i fi ed 73 g en e-b as ed da t a s e t s with S N P -g ene a s soci atio n s (mol ecula r QTL s , promo t e r- enha nce r inter ac t i on s) or g ene -ce n t ric ( dif fe ren tial e xpres s i on, GWAS t arge t p r i or i ti sati on p i peline s – Pr i xFix e, regfm) in fo rmatio n (Figure 1 s ub p anel 2 , Addition al fil e 2: Supplem e nta r y Fig ur e 1 ). Prioritisation of candidate genes ADGAP P gene rat e s a c au sal g en e priori ti sa tion score for ev er y gen e within a 3M b w indow c entred on e ach o f t he 25 in dex G WAS SN P s and a ca us al S NP p riori ti s a t i on s c o re for eve ry SNP wi t hi n th e LD-ge n era ted bou nda r y , whi ch a llow s su bs e q uen t ranking of gen e s a nd va r ia n ts a t e ach locu s (Fig ure 1 subpan el 1) . To g enera te the p rioriti sa tion scor e, each s ou rc e of ev idenc e is a s s ig ned a we ight ba sed on s ubj ective streng th of e videnc e: hi ghe st f o r r e sult s f rom st ati st i c al t e st s u s ing fu ll se t of s u mmary sta ti s tic s , suc h a s molec u lar QTL c oloc alisa tion me thod s; low es t f o r predi cti on re sult s from mac hine le arning model s su ch as v arian t func tion al p redic ti on s of t w are and interm edia te fo r po s i t i on al ov er l ap wi t h s ig nific a nt expe rime nta l re s u l ts, such a s identi fied promot er -enha nc er loop s (Fig ur e 1 subp anel 3 & 4) . In c alcu la ting the f i nal scor e, we al s o took into ac count th e magnitude of re s ult signi fic a nc e or e ff ec t, s pe ci fic ity (ove r a ll numbe r o f SN P s /ge ne s signific ant in a giv en ex perim ent ), inde p e ndenc e o f ev idenc e (number of ex pe riment s condu ct ed in the same stu dy, s uc h a s mea su ring bo t h expre ss i o n and DNA me thyla t i on lev el s ) . T he fina l s c or e wa s adju s t ed by het eroge ne ity o f ev idenc e (I .e. g ene s o r varia nts c on s i s ten tly suppo rt ed by a rang e of ev idenc e s ou r c e s - alt ern ative f unction al as s a y s and sta t i s tic al metho ds – a r e upw ei ghted in propor tion t o the squ are r oot o f mea n n umber of u ni que study typ e s a nd unique s tu dy IDs) , a s w ell as ab sol ute nu mb er o f studi e s p rov iding suppo rtive evid e nce, con si s t en t w ith cri t eria u s ed in triangul a tion and a ss e ss m e nt o f cau sal li nks in e pid emiol ogy, suc h a s B rad ford - Hil l criteria [143 ]. Gene p riori ti sati on sco re s ra nged f r om 0 to 1405 while for S NP s from 0.5 t o 968 ( Dat as et S4 ). Fo r 8 loc i the top p riori ti sed S NP w a s not the i ndex SNP, a nd for 10 lo ci t he c lo se s t g e n e did not s c or e be st (Tabl e 1) . In de tailing t he re sul ts, w e f ocu s on gen e s ranke d in t he t op 3 and S NP s rank e d in top 1 0 a t eac h loc u s as although qui te a rbit rary, th is limi t agr ee s with th e sharp sco re deca y obse rved i n ADGAP P s c or e s (Addi tion al Fi l e 3 & 4) . Ex cludi ng the c omplex MH C locus , the h i ghes t gene s c ore s w e re se en fo r gen e s a t 5 loc i: IL1 8R1 (sco re =138 4) and IL 18RAP (sc ore = 1 341) at th e 2 q12.1 l ocu s, P PP2 R3 C ( s c ore = 99 6) a t t h e 14q13 .2 loc us , IL7R ( score = 965) a t the 5p13 .2 l ocus, TRAF 3 ( s c ore = 848) a t 14q32.3 2 loc u s a nd IL6 R (s co r e = 743) at 1q21.3 locu s (Ta b le 1) . A s s umi ng that the t rue mod e l is on e of a singl e c au sal g en e a t e ach loc us (un likel y to al way s be true ), pr i o r iti sa tion c an a l s o b e ev aluat ed b y co mpari ng the s c o r e o f th e top prio riti sed ge ne at a l ocus wit h a ll ot her ge ne s a t tha t loc u s . E ight loc i (1q21 . 3 - IL6 R , 10q21.2 - AD O , 11p13 - P RR5L , 5 p13.2 - IL 7R , 1 1q 24 .3 - ET S 1 , 2 q37.1 - IN PP 5D , 12q15 - MD M1 , 14 q32.32 - TRAF3 ; Table 1) h av e a sing l e s t a nd -ou t can didat e fo r c au s a l gen e – w it h th e top g ene cont ributing more than 50% o f t he tota l sco re o f top 10 - ran ke d g ene s. The top c andi d ate by tha t met r ic is PRR5L (79% of top 10 g en e s at 11p1 3 loc u s ) , w ith a sco re o f 598 c ompared to 65 fo r t he s e con d- ranke d gen e at t hi s loc us. Mo st top -pr i o r i t i s ed g ene s by the to t a l s c or e ar e a l s o pr i o r iti se d by this met ric. . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint Tw o furt h e r l oci s how good e videnc e (>7 5% cu mulative s c or e) s ha re d ac r o s s t w o ca ndidat e g ene s ( IL 18R1 and IL18RA P a t 2q12.1 a nd EMS Y and LRRC32 at 11q13.5 , whic h sh are 77 % and 8 4% of the cu mulative s c o re r e s pe ctive ly ). A t 2q12. 1 (where I L18R1 and IL 18RAP re side ) th er e is ev ide nc e fo r two i ndepen de nt ge netic s i gnal s, a nd t h e s e may a ffe ct eac h o f th e se ge n es. Out o f gene s a t the s e 11 loci with g ood prioriti sa t i on evi dence , 6 hav e strong s up p or tiv e e QTL co local isa t i on evi dence (c oloc po st e r io r proba bili ty of coloc ali sa tion (P P H 4) > 90% or T W AS p -value <1 x 10 -5 , Figu r e 2, and A d dition al F ile 5 : Supple men tary Ta ble 1 ): PRR 5 L i n whole blood, LCL , s k in and CD 4 + T c ells , TRAF3 in w hole blo o d, IL 6R in wh ole bloo d, IL1 8R1 in w hole b loo d and LCL , IL1 8RAP in w hole b lood and CD 4 + ce ll s , LRRC32 in wh ole bloo d, P PP2 R3 C in skin, C D 15 + gr anulo cyte s, an d who le blood . In add iti on, 20 oth e r g en es ranked 1-3 in AD GA PP hav e strong coloc alisa tion e videnc e. For the majori ty of the loci, we pri oriti s e ge ne s t ha t have pr e viou sly b e en c on s id e r e d in an no t a ti on of th e se G WA S loc i [11] , but A DGAP P pro vides addi t i onal re solu tion on the like ly c aus al ge ne s through th e i nte grati on of n e w ev idenc e . Thus, f or many loci we are abl e t o s t ren gthen t h e evid ence for ex i s ting cand id at es. Howeve r, f o r so me l oci, our pr e sent a n alysi s pri ori tise s a n altern ative gen e to the origi nal G WAS . At 10p1 5.1 IL 15RA was prev iou s l y re por ted a s th e mo st lik el y c andida t e . O u r ADGAP P analy s i s rank s IL 2RA in the top spot ( scor e=3 31 ). Sub se quen t to t he EAG L E GWAS, CRIS PR ex periment s h ave been c ond ucted whic h hav e es t a bli sh ed tha t th e S N P whic h A D G AP P r a n ks a s top SNP a t thi s loc u s, re gulat e s IL 2RA exp r e s sion, p rovid ing v alidation for o ur pr i o r iti s ation approa ch [144 ]. In a dditi on, f or fiv e loc i, AD G A PP prio r i tis es gen e s in t o p po si t i on ( and w ith a scor e > 300) th at w ere not c on s id ere d in th e o r ig inal an nota tion of the s e lo ci in t h e GW A S pape r[11 ]; MDM 1 at 12q 15 (sco re =728 ) , AD O a t 10 q21.2 ( scor e=6 15 ), STMN3 a t 20q13.33 (sc or e = 608), S LC22A5 at 5q31.1 (sco re =461 ) and DEXI a t 16p13.1 3 ( s c ore =376). S om e in t hi s l i s t ( such a s SL C 2 2A 5) r e pr e sent promising l o oking cand idat es , whil st oth er s are in l oci w he re th ere are fa r more p r o mis i ng biol o gica l ca ndidat es (e .g. at 12q15 , whe re IL22 a nd IFNG ap pe a r t o re pr e sent mor e p lau sib le c andida te s than the to p rank ing MD M1 ) . It w ill th ere for e be of in ter e s t t o pro spec tiv ely f o llow th e se l oci, a s m o r e ev idenc e i s gathe red a n d ex perimen tal w ork done, to a sse s s wheth er any o f the n ov el c andida te s pre sen ted he re do in fac t t ran spi re to b e t h e ca u s a l t arge t gene s . In orde r to i nd epend ently e s t a bli sh if re stric t i on o f t he ti s s u es to th os e kno wn to be mecha ni s tic ally link ed to ec ze ma and en riched in our G WA S si gnal in MA G MA a nd S NPS ea an al ys i s w as lik e ly to ha ve influe nc ed t h e final g en e rank ing, w e c o mpared our re sult s to tho s e run on the f ull se t of 53 G TE x tis s u e s, Blu e pr i n t and e QTL Gen dat as ets using Mend elia n r andomiza tion -ba se d c oloc ali satio n method, SM R. Tw elve out o f 20 (A dditio na l Fil e 6: S upplem en ta r y Table 2) ge n e s show ing s tr o ng ev idenc e (p <6.5 x 10 -6 ) f o r c oloc al is ation using SMR wer e among th e t op 3 hit s for a giv en loc us from our AD G A PP a naly s i s . O f the remai ning 8 , 3 were pr i or i ti sed by SMR in ti ssu e s ab s ent in A DGAP P : th e es tabli sh ed ec z e ma fila ggrin g en e in a ort a arte r y (rank ed 6 at 1q21 .3 - a l ocu s in A DGAP P), i t s anti sen s e tra ns c ript FL G-AS1 in e soph agu s mu scu lari s ( r a nked 21 a t the sam e lo c us ), and RT EL1 heli ca se in t i bi al a nd aor ta a rte ry (rank ed 15 a t t he 20q13.33 loc u s). How eve r, fu rt her inve s tiga t i on of th e se e QTL s ig n al s s u gge s t some may not be biol ogic ally meaning ful du e to ver y low ex pr e ssion lev els o f th e se gen e s i n t h e r e s pe ctive t i ssue s (FLG TP M 0 .59 in aorta ar tery and F LG-AS1 T P M 0.62 i n es ophag u s musc ul ari s in GTEx [ 2 0 ]) and s o inc lusion o f lik ely ir r el evant ti s s ue s ma y inc reas e the ch ance of s pu r i ou s re sul t s . Ther ef ore , re stric t i on to s i x GTEx tissue s i s unlik ely to ha ve h ad signi fic an t impa ct on the fi nal ge n e ranki ng. On bala nce, r e s t riction o f t i ssu e s will r e t a i n e QTLs shar ed ac ro ss tis s u e s[14 ] bu t remov e s t h e probl em of i ncrea sed fal se p o s i tiv e ra te for e QTL s in tis s u e s l e ss rel ated t o t h e pat h oph e n ot y p e[ 1 4 5] . . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint Prioritisation results for choice loci A full di s c u ssion o f each locu s in Table 1 i s ava ilabl e in Addit iona l Fi le 7 : S uppleme n t a r y Tex t 2. He re we highli ght seve ral loci with e sp ec ially compe lling pr i or i ti sati on e viden ce (F igure 2 & 3, s ee Da t a set S3 and Additio na l Fi le 8: S upp leme n t a r y Fig ur e 4 for all loci) and in teg r a te t hi s w it h know ledge fr om litera tur e. Locus 1q21.3 - b At thi s locu s we t ested 104 ge n e s within the 3 Mb int erval (Figu re 3A & Da ta se t S 4 ) . In terl eukin 6 r e c ept or ( IL 6R) had t he highe st prio riti s a t i on s c ore – 7 43 (62% of top 10 c umula ti v e score ; Ta ble 1 & Fi gure 3A) , wi t h t h e n ex t b e s t g en e bei ng UBE2Q1 - s c or e of 93 (on ly ad diti onal 8 % of top 1 0 cu mulative s c o re ). The high prio riti sa tion of IL 6R by ADGA PP i s mo stly driv en by coloc ali satio n (Fig ure 2 ) in w hole b lood in e Q TL G en an d GTe x (c oloc p o s te rior p robabil i t y of c ol ocal isa t i on (P PH4) = 63% -88% a nd TWAS p -v alue = 5 x 10 -6 ), in lipopoly s a cc ha ride - / mu r a myl dip ep tid e- prim ed monoc yte s in Kim - Hell mu th et al . (P PH 4 = 75% -77%) a nd s un -un expo sed skin in G TEx (PPH4 = 64% ) (Addi tiona l Fil e 5: Su pplem enta ry Table 1[11] . Po s s i bl e pla ceme nt o f many l ocus i nterval SN P s within an enhanc e r of the ge n e is highl igh ted b y H i C inter actio n da ta in human e mb r y o nic s t em c ells [79 ], who le blood [8 1] , C D34+ he ma topoi etic c ell s a nd lymph obla stoid c e ll line s[146] an d epi dermal s tem ce lls alon g with k e ratin ocy te s[84] . Pre viously, IL6R w a s al so impli cat ed a s th e po tenti al c au sal g ene at thi s loc u s i n th e EAGLE G W AS [11] , as one of the varia nt s in th e MANTR A -ba se d c r ed ible se t of 23 va r i ant s , r s222814 5, r epr ese nts a mi ssen se (A s p 358Al a) mut atio n in the g en e [14 7]. Thi s varia n t is ranke d 3 rd in A D GA PP and is ass oci at ed wi t h blood s erum lev el s o f IL6 R pro t e in a s we ll a s tran s c rip t [127] . Howev er, the ac t ual ac ti on of r s2228145 i s t o s hi ft IL 6R pro tein fr om mem br a ne - bound sta te to sol uble cel l fr a cti on, or in other wo rd s, f rom the a nti -inf la mmato ry c las s ic al signa lling pathw ay to pro -in fla mma tory tr an s - signa lling [148] . Mend el ian ra ndom iz a tion ana ly s e s in dica ted inc r e a s e d s o lub le IL 6R le vel s a s ca u s a l fo r higher A D (and a sthma ) r i sk and dec re a sed solu ble IL 6R lev els indic ative o f increa sed r he u matoid arthri ti s ( R A) ri sk [149] . This i s in ag ree m ent with th e Asp358 Ala m ut a t i on be ing prote ctive to w ards R A and in cre a sing risk in A D[150 ]. H ow eve r , in our c ol o c a li s at io n a n a ly s e s t he A D r is k al le l e is l i n k e d t o o v e r a l l r ed u c t io n in IL6R t ra ns cript l evel s, li kely due t o t h e compl ex ba lanc e o f IL 6 cl a ssic a l and t ra ns - s i gnall ing . A rang e o f an ti- IL 6 biol ogics ar e in cl inica l use a nd devel opm ent for in flam ma t o r y di sea s es, includ ing, toc ilizum ab, w hich inhibits bo th solubl e and membr an e -bound IL 6R, and i s a ppr oved fo r tr ea tment o f RA a n d juve nile i diopa thic arthri ti s[151 ]. H ow eve r , AD a dve r s e eve n t s h av e bee n r e po r te d in toci liz u mab tri als [152 ], c on s i ste nt wi t h t he opp osi ng e ffec ts o f t hi s GWA S loc us on R A and AD. Locus 2q12.1 This loc u s con sis ts o f two indep end en t s i g nals r epre se nte d by rs641957 3 and rs3 917265 lead SNP s [11] . Alt ogeth er, w e co n s id e r e d 46 ge ne s loca ted within t h e 3 Mbp int erva l of ea ch S N P (F igure 3B & Da ta se t S4) . A s menti one d e arlie r, t he highe st s c or es we re ob tai ne d by Inter leuki n -18 rece ptor 1 - I L18R1 (sco r e = 1384, T able 1 ) and I nte rle ukin-18 r e cep t or ac ce ssory pro tein 1 - IL 18RAP (sco re =134 1) whic h toge the r a ccou nt f o r 7 7% of the t op 10 c umula tive s c o re a t th e lo cus , and so a re lik ely t o r ep re sen t the c au s al g ene s behi nd t he two signal s . I L18R1 and IL 18RA P ’s r ol es a r e suppo rte d by a bundan t ex pre ssion c oloc alisa tion ev idenc e in th e whole blood in G T Ex ( IL 18R1 : PP H 4 = 5 1%-80% and p-v alue = 7 x 10 -6 ; IL 18RAP : P P H 4 = 96% - 9 9% a nd p -val ue = 4 x 10 -12 ) a nd immune cel l type s in CED A R, Twin sUK and Kim-Hellm uth et a l . ( IL18 R1 : P PH4 = 55%-86% and p-v alue = 2 x 10 -5 , IL 18RAP : PPH4 = 55%-97% and p-valu e s = 5 x 1 0 -5 to 6 x 10 -8 ) , a s w e l l a s sk i n i n T w i n s UK fo r IL 18R1 only (p-val u e = 1 x 10 -4 ; Figure 2 & A d diti onal F ile 5 : Suppl e ment ary Ta ble 1) . T he direc t i on of e f fec t indi cate s inc rea s e d expr e s s i on o f both g e ne s fo r A D ri sk al lele s, wi th one ex cep ti on in GTEx whole . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint blood fo r IL18RA P (Add i t i on al F ile 5: Sup pleme nta r y Tab le 1) . Int er es tingly , oppo s ing e xpres s io n ef fec ts o f th e s ame alle l e in a varian t h a s p r ev iously be en re por ted in mo nocyt es and whole blood for IL1 8R AP [104] . SN P s in L D with th e l ea d S N P a t t h e thr e e top -ra nked gen es ove r la pped w hol e blood mQ T L SN P s in the Go DMC s t u dy[1 26], l ikely influ enc ing t he me thyla t i on of promot er s of tho se gen e s . D i f fer enti a l gene expr e ssi on ( DG E ) consi sten tly s h ow s s ig nifica nt upr egulat ion of IL 18RAP [132] an d IL18 R1 [135 ,153 ] in the skin of A D p a tien t s ( Dat as et S5 ). Bo t h ge nes w e r e put forwar d a s c a ndida te gene s in the EA GLE GWAS an n ot a t i on, as the c re dible se t va r ia nt s span b o t h gen e s a nd indi vidual varia n ts link ed t o IL 18R1 e Q TL s in Twi ns UK skin microarr ay da ta, bu t previ ou s ly ther e w as li tt l e evi dence for coloc ali s atio n. In g ener al, c ytoki ne IL - 1 8, in th e f o rma t i o n of who se r ec ept or compl ex both I L18RAP a nd IL 18R1 partic ipa te , ha s bee n s h own to el icit sign ifi cant im mune ge n e ex pr e s sio n c hange i n keratino cyt e s from A D le sion s [ 1 53] . IL 18R1 and IL18 RA P are bo th inv olved in T-c ell , e s pec iall y T he lper ce ll signal ling a nd ac t i vation o f t h e N F kB -pa t hwa y. IL 18R1 a nd IL1 8RAP medi a te IL-18 -de p enden t signa l tran s d uc tion, w hich is o f s pe cial impo rta nce in Th1 r e s p on se . T hus , i t would not b e su r p r i sing if on e va r i ant a ff ect ed t he expre s sion o f the w h ole g ene c lu s t er, a nd tha t ha s inde e d bee n s how n to b e th e ca s e with S N P r s917997 a s s oc ia ted w ith I BD and c oeli a c dis ea se [154 ]; and r anke d 8 th am ong the tw o signal s . A v ariant si tuat ed 15 kbp aw ay from rs91799 7 - r s 990 171, a gain a ssocia te d with c elia c dise a se an d lympho cyte c oun t s[14 0], i s ra nked a s sec ond -be s t among the t w o sig nal s in t h e loc u s ( D at as et S 4 ). T he r e is s tr o n g pQ T L s u ppor t f or ass oc i a t io n o f r s 9901 71 wi t h IL18R and IL 1RL1 re sulting in incre a sed p rotei n avai labili t y [128] (D ata s et S5) . I n add ition , rs99017 1 overla p s a llel e - speci fi c e QTL s for IL 18RAP a nd ace t y la tio n hQTL s i n n eut rophil s [21] a ppea r i ng dur ing Th1 p o l ar iz at i o n[ 1 55] . Locus 5p13.2 We c on s id ere d 41 pot enti a l ca ndida te g e nes s i t ua te d within the 3 Mbp w indow centr ed on index SNP for thi s loc u s (F igure 3 C & Dat as et S 4). Mo s t of the c umulativ e sco re wa s a s si gne d to in terleuk i n- 7 rec eptor sub unit alp ha - I L7 R ( s c ore =96 5, whic h contribut e s to 65% of t op 10 c umula t i ve sco re, Tab le 1), foll owed by SP EF2 ( scor e=2 03 a nd a fur t h er 14%) . Whil e our G W AS re s ul ts d o not di rec tly co local ise with a ny e QTLs for the g en e (i n da ta s e ts whe r e thi s c ould b e t e s t ed), the r e a re multipl e e QTL assoc ia tion s f or SNP s in LD wi t h t h e index SNP: in whole blo od [117,1 21] , C D 4+ T ce lls [ 2 1,102,10 3 ], mac r op hage s [108] an d monocy te s [10 1 ] and p QTL s in whole bl ood[127, 128 ] a s we ll as pr omote r- enhanc er int erac tion s i n human em br y onic stem c ell s [79], CD3 4+ hemato poie t i c ce lls [ 1 46 ], naï ve T r eg ula tory c ell s and T helpe r 1 7 ce ll s[15 6] (Figure 2, Da ta se t S5 ). IL 7R i s among the gen e s fou n d to be up reg ula ted in th e s k in in eczema p ati ent s i n a me ta -a naly si s [1 32]. Ou r r e s ul t s co nfir m initi al pr i or i ti sa tion o f IL7R i n the GW AS whic h wa s supp ort ed by 18 c r e di ble s et va r i an t s spanni ng the gen e , inc luding one no n- sy nony mous (Th r > Il e) va r i ant, r s 689 7932[1 1]. In ADGA PP, rs 6 897932 wa s ranked a s t h e 8 th mo s t like ly c ausal v ari ant . The variant a f f e c ts s pli c ing of the IL 7R tran s c rip t, with th e minor C al lele , th e r i s k a llele fo r multiple sc le ro s i s (MS ), fav ou ring secr et ed over surfa ce i so fo rm of th e pro tein . Elev at ed lev els o f s e cr et ed is ofo rm exa cerba te sy mptoms o f MS in ani mal mode l [157,158 ]. In co mmon with o ppos i te e f fect s s e e n in A D c ompare d to autoim mune dise a se s, t he minor a ll ele i s p r o tec ti ve in eczema [15 0] . IL7 R is p art o f th e thymic stromal ly mphopoietin (T SLP) rec ep tor / IL -7 / I L7R a xis requi red for c orrec t lymphoc yte ma turation , e s p ecia l ly of Th2 lympho cyte s of in ter e st in A D, wi th overexp r e s sion a s s o cia ted wit h ac ute l ym phoblas tic leuk aemia [159] , wh er e a s r e ce s s iv e mut a tion s in the ge ne r e sul ting in reduc t i on o f gene exp re ssi on is se en in sev er e com bined im munode fic ie ncy (SCI D) pa tie n ts[160 ]. Locus 11p13 . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint This loc u s stand s ou t a s ha ving one c lea rl y p r ioriti sed targe t g ene - PR R5 L (sco re = 59 8, 79% of top 10 cu mulative s c o re; Tab l e 1), w ith i t s fi n al s core ni ne t ime s highe r tha n th e s econd -r ank ed ge ne, TRAF6 ( s c ore =65 - on ly 9% ). In to tal, w e c ons id e r e d 15 g en e s loca te d in the 3 Mb p region arou nd inde x SNP (F igu re 3D & D ata se t S4) . We c ould not t ea se a p ar t t h e targe t behin d th e se conda r y s i gnal repr ese nt ed by lead S NP rs12295 535, w ith the sam e g ene s prio r i t i s e d a s fo r the p r im ary signal repr ese nt ed by r s25925 55 . r s 25 92555, the top pri oriti sed varian t a t the loc u s ( sc ore=24 6 , Da ta se t S4), i s s i tua ted in th e int ron o f prol i ne ric h 5 lik e (PRR5L ) . We als o not e t ha t ADGA PP’ s s e co nd be s t SNP r s7925585 ( s c ore =132 ) fo r th e prim a r y s i gnal, al so po s i t i on ed wi t h in t he int r on of P RR5L , w as inde pend ently pr i o r i t i s ed fo r ec z e ma in F INDOR an a lysi s[16 1] . Str ong e xpre s s ion coloc alis ation w a s detec t e d with the coloc met hod (F igu re 2, A d dition al F ile 5 : Suppl e men tary Ta ble 1) i n: e QTLGen - who le blood ( PPH4 = 95% ); Twi ns UK – s k i n (P PH4 = 91 %), L CL ( PPH4 = 98%); C E D A R - C D 4 + T c ell s (P PH4 = 98%) w ith pro t e c tive AD all ele a s s oc ia ted w ith inc rea s ed expre s s io n. The g ene’ s role w a s also supp ort ed u sing the n etwork me tho d Pr i xFi xe (Da ta se t S5) a n d strong e st met hy lation signa l detec t e d t hr o ugh m QTL overlap with loc us int er v al S NP s in whole blood in t h e Go D MC s tudy[126 ]. PRR5L wa s pr opo sed a s the candi d at e ge ne in the E AGL E GW AS du e to p osi tion o f the lead S N P in PRR5L ’ s intron and PR R5L e QTL overlap with the c re dible se t v ariant s[11 ]. H ow ev er, th ere wa s previ ou s ly very little ev idenc e f or coloc al isa tion o f t he se s ig n al s. PRR5L is pa rt o f th e rapam ycin compl ex 2 ( m TO RC2 ) whic h re s po nd s to e xtrin s ic st i muli through cy tos k el eton re -organi s ation a nd c ell mig r a tion [162 ]. PRR5L sp eci f i call y pla y s a rol e in r e gulati on o f fibrobl a st migrat ion , and d e crea s ed ex pre ssio n of the gene con f err ed by t he ri s k allel e i s pre dicted to lea d to inc r e a se in fib robla s t migratio n. Locus 14q13.2 Out o f 70 ge ne s con side red a s c au sal a t t his loc u s (Figu re 3 E & Da t a set S 4 ), we fin d 2 of the m t o hav e co mparably high sco r e s - P r ot e i n pho s p h at ase 2 r e gu lat or y subun i t B ' ' G a mma ( P PP2 R3C) w ith the score o f 996 (31% of top 10 c umula t i v e score , T able 1) , and KI AA03 91 with t h e s c or e o f 814 (25% of top 10 cumul a tive sco re ). Three (P P2 R3C, KI AA03 91, FA M177 A 1 ) o ut of the f our t op- r a nking A D GA P P g ene s at thi s locu s display pa rtial c o -expr es si on. The t op - ra nk ed c andida te gene PP P2R3C shows c ol oca lis ation (Fig ure 2, Addi t i on al F ile 5: Suppl em en tary Tab le 1) in s un -ex po sed (P PH4 = 94%) an d un e xpo s ed s k in ( PPH4 = 9 5%, p -v alue = 2 x 10 -7 ), w hole bl ood in GTE x (PPH4 = 97%, p -value = 2 x 10 -7 ), C D 15 + granulo cyte s (P PH4 = 96%) a nd col on (P PH4 = 96 %) in CE D A R, and neu tro phil s ( PPH4 = 93%) i n Blueprin t. Th e n ex t bes t gene , KIAA0391 r ec apitul a tes the c o loc alis ation in t h e s k in in GTE x (su n ex po s ed PPH4 = 97 %, unex pose d P PH4 = 96%), a n d LCL and ind ivid ual immu ne ce ll type s: L CL f rom Tw insU K ( PPH4 = 94% , p -va lu e = 3 x 1 0 -9 ), CD8+ T cell s ( PPH4 = 97%) and CD14 + monocy t e s from CE DAR c ohort (P PH 4 = 60% , p -va lue = 1 x 10 -4 ), in a dditi on to t h e spl een in GTEx (P PH4 = 95 % and p -valu e = 2 x 10 -7 ). Moreove r, w e fin d hundre ds o f indivi dua l v ariant s in L D w ith th e ind ex S N P to ove rlap b lood and s k in e QTLs f or tho s e gene s ( Da ta se t S5) . T he orche s t rat ed ex pr e ssion o f th e se g en e s is unde r s c ored by their di f fer entia l expre s s i on in a t op ic de rma t i t i s skin: P PP2 R3C i s up reg ula ted r ega r dl e ss o f FL G gen ot y pe [135,1 37 ] and K IAA0 391 is stro ngl y dow nr e gulated in homo zygou s FL G mutation AD patie n ts[135 ] . Com pa ring tha t w ith our c olo cali s atio n r e sul ts , we see ti ssu e- sp ecif ic regula t i on for PP P2 R3C, upr egul ati on in the s k in (in li n e with the dif fe ren tial e xpr essi on r e sul t s) but dow nr e gulati on in the bl oo d a s s oc ia ted w ith the A D ri s k alle l e. For KI AA03 91, t he dow nregula ted ex pr e s si on seen i n AD pa tien t s is a t odd s with the e Q TL re s ul t , wher e the A D ri sk a llele i s a ssocia t e d wi t h increa sed e xpre s s ion; t hi s c onflic t could indica te tha t KIAA0391 is n ot an A D s us c ept ib i l i t y ge n e. The or i ginal GWAS anno ta tion [11] a lso s ugg es t ed P PP2 R3C, KIAA0391 and FAM1 77A1 as p l au s ib l e ca us a l gen e s , with th e 47 credibl e inte rval SNP s sca tte red t h roug hou t the thre e g en e s a nd th e le ad . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint SNP ma pping t o a n intr on withi n P PP2 R3 C ; s t rong c oloca li sa tion wi t h Twins UK mic r oa r r ay eQTL s in that pa pe r wa s f ound onl y fo r KIAA0391 . Con side ring gene fun ction , PP P2R3C i s th e mos t via ble c andida t e . Target ed B - c ell mous e knoc kou t mutant s s ho w pro found a bnormali tie s in hum or a l immune r e s po ns e: r educ ed B c ell pr ol i fera tion , matura tion, ab normal a c tiva t i on an d sm all s pl een [163] . S imila r ly , lo s s o f P PP2R3 C i n T c ells re s u lts in atrophy o f t h e thymus , dec re a s e d t hymoc yte abun dance , e specia lly o f C D 4 + an d C D 8 + double - posi tive t hy mocy t e s[164] . V arian t s a t th e locu s rel a te d to pl eio tropic pheno typ e o f ch r oni c infla mmat ory dis ea se s (AS, CD , ps oria si s , p r im ary scle ro sing c holang iti s , UC) [165 ] map to intr on posi tion s within P P P2R3C. N ex t , linki ng t he g ene’ s mut ant ph enoty p e e stabl is hed in mous e to huma n G WA S, it ha s be en re po rted th a t d ecrea s e in lymphoc yte c ount s[166 ] s ig n ifica ntly c orrela t e s wi t h t he ri s k alle l e at r s2038255 , the ind ex va r ia nt in AD GW A S and A DGAP P top prioriti se d varia n t (sco re =389 , Da ta s e t S 4) . P PP2 R3C enc od es a r egul a tor y sub unit o f pro tein ph o sp hata se 2 A, k nown as G5PR, w hich a ssoci a te s with pho sph at as es PP2 A, P P5 , G A N P pro tein an d repr e s s e s J NK and IK K β ( i n h i bi t o r of N F- κ B) pho sph or y lati on [167 ] . I t i s involv ed in r e gula ting antigen -ba se d B-cell and ea r l y T ce ll sel ec tion in the thy mus pr o m ot i ng th ymoc yte a nd B ce ll su rviva l. G5P R bec om es up regula t e d in ac tivated B c ell s a nd preven t s B -cel l rec e p t or -media t e d ac t i vation -induc ed c ell d e ath in B cell s through suppr e ssion o f lat e-ph a se J NK ac t i vation [168 ] . O v e rexpre s s ion re s ult s in t he increa s e of produc t io n of non - sp ecific B c ell s a ft er i mmunisat ion and ge n era tion o f aut oa nt i bodie s in non - s t im u la t ed m ic e [ 1 6 9] . T h er ef o re , P PP2 R3 C upregul ati on in the s k in cou ld be a con tributi ng f a ct or to autoimm une ac t i vation s een in a sub se t of sever e AD p ati en ts and pa rt i cula r ly dir ected agai n st epid er m al p rot ein s[170 ] . KIAA0391 in c ontra st do es n ot app ear to be so di rec t l y func t i on ally linke d to th e A D pheno type . It enc ode s a c ompon ent o f mi t oc hon drial RN a se P comple x w hich cataly s e s the l a st s tep in pre -tRN A matura tion p r oc e ss: remov al o f th e t R NA 5’ l eader se quenc e [171 ]. Locus 14q32.32 Acc ounting f o r 55 % of th e cumula t i ve sc ore a t the locu s, TNF r e cep tor a ss ocia ted factor 3 (TRA F 3) i s cl early p r io r i ti sed a mong t he 5 9 gene s p osi tioned wit hin 3 Mbp of the in d ex S N P a t th e loc us (F igure 3F & Da ta se t S4 ). W hi le TRAF 3 ’ s sc o r e i s 8 48, the sec ond -ra nked AM N score s only 281 (18% ) (Table 1). TR A F3 i s th e only gene at the lo cu s wi th direc t colo cali sati on evide nce (Addi tio n al Fi le 5: Sup plemen tary Table 1 ): in th e w hole bl o od in e Q TL G e n ( P PH4 = 9 3%) and with lo w er confi d ence (P PH4 = 85%) in C D 4 + T c ell s in Bl uep rint . In add ition , many loc us i n t e r v al S N P s ar e po ssibly s itu at ed wi t hi n an e nhanc e r inte r a c ting t he gen e’ s promo te r in h uman emb r y onic st em c ells [79] , wh ole blood [ 8 1 ], CD34 + he ma topoi etic cell s a n d lymph oblas toid ce ll lin e s[146 ], naïve T regul at or y ce lls and T he lper 17 c ell s[156 ] a nd epi d er m al s t e m cel ls a long w ith ke ra t in ocyt e s[84 ] a s s hown by Hi-C da ta. In AD GWAS , ri sk all ele s cor rela te with u pregul at ed ex pr e ssio n of T RA F 3 a nd in IB D the re i s inc r e a s e d exp r e s sion o f the g e ne in infl a med inte s t ina l muc o s a [17 2] . Howeve r , c han ge s in ex pr e s si on ha ve no t be en c on s i s ten tly ob s e rve d in A D le sion s rel ative t o heal thy s k in ( Dat a set S5 ). The or i ginal EAGLE G WAS ann ot a t i on al s o pre sen t s TR A F3 a s th e c andida t e gene at th e loc us d u e to the lo ca tion o f t he ind ex SNP, p athw ay e nr i chment i n MAGENTA g e ne se t anal y s i s a nd mous e knoc kout ph eno type, bu t no e QTL e vide nc e[11 ]. T wo out o f th e 3 top pri oriti se d SNP s a t the loc u s i n ADGAP P (1 st rank ed r s 79 589176 and 3 rd r ank ed r s 1 2880641 ) a s well a s index SN P ( rs 714 6581 ) are intronic , s i tua te d within t h e TRAF3 gene , w herea s th e remai ning top 3 SN P (2 nd r a nked rs714212 62 ) is 2 k bp 5’ ups tr eam of TR A F3 . TRAF 3’ s role in signa l tr a n s d uc tion in immunity i s w ell-e s tabli she d, wi t h early s tudi es d e scribing a seri ou s imbal ance in T c ell compo sition in mou se model k nock out s whi ch e ventuall y lea d s t o the i r perin atal de ath [173 ]. TRAF 3 i s a r epre s s o r of CD40 - a nd B c ell - . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint ac tivating f actor -medi at ed s ig n allin g and l imits home o st a t i c B c ell s urviv al[174 ]. TR A F 3 is al so rela t e d to po s s i ble c andid at es at two o ther G W A S l oci ( TR A F 6 at 11 p13 and IL6R a t 1q21.3) . TRAF6 posi tivel y regula t e s MA PK sig nalli ng and produc tion of in fl ammat ory c ytokine s a n d che mokines , whe r e a s TRAF3 n e ed s t o b e deg r a d ativel y ubiqutina te d during My D 88 - d ep ende nt To ll-like rec epto r signal ling to a ctiva te the JN K and p38 M A P K ca s c ade [175] . T RAF3 ex er ts a nega ti v e ef fec t on Th17 - bas ed i nfl ammation by sequ e ste r i ng IL -1 7R, how ever, no rmal ly thi s is p reven ted by c ompetitive bindi ng of TRAF 3 by N R D1. T hi s allow s fo rmation o f IL17R -Ac t 1 - T RAF6 c omplex an d sub seq uent propag at ion o f IL -17 -induc ed signal dow n through M A PK and NF -κB pa thway s l ea ding to producti on of pro -in flam mat or y molec ule s, inc ludi n g cytok ine IL -6, who s e r ec ept or i s prior i t i sed in the A D GWA S a t loc u s 1q21.3 [176 ]. Validation of ADGAPP gene prioritisation – enrichment and network analysis Due t o li mited know ledg e of h erit able a t opic dermati ti s loci, we di d not h av e any s ig nific a nt numbe r of “ gold s tanda rd” tr ue po s i t i ve ge ne s t o whic h we could compa r e our r a nking o f prioriti sed gene s in the a topi c derma titi s GWA S . For that r ea son, w e ev aluat ed t h e quality o f our r e s u l t s in tw o indirec t wa ys. Fi r s tly, ma ny GWAS ge n e prio riti s a t i on al go r ithm s foc u s on pr i or i tisi ng gene s wh ich shar e simil ar profil e s, be it in memb e rship i n gen e set s , co-e xpre s s i on or pr o tei n - p rot ein int era ction netwo r k s[177] . H er e, we w ork thi s appr o ac h in reve rse a n d a sk if ou r p r i oriti s ed g ene s ar e enric h ed for t heir pr e senc e in any gene s et s an d n etwork s. Using en richr [178], we c a rr i ed out ge ne se t enric hmen t te sts of ou r top 3 pr i o r itiz ed g ene s, to s e e if they alig n wel l with c at egorie s f rom addi t i on al prev iou sly impl icat ed A D g en e s ( Additi onal File 9: Sup pl emen tary T able 3, Addition al File 10: Supplem en tary Tab le 4 ). In gen eral , h ighly -pr i or i ti sed g en e s had fun c tion s re la ted c hie fly t o t h e immune s y s tem bu t al so de rmis s t ructu r e, li pid meta bo li s m and c yto s k ele ton org a nisa t i on . We find tha t bot h li sts a re signi fic antly e nrich ed f or immune s ys tem -rel ate d g ene s (Fig u r e 4) . In partic ula r, cy tokine ca tegori e s w ere ove r repr es ent ed: GO cytoki ne -medi at ed s ig n allin g pathw ay (adju sted p - va l ue f or AD G A PP p r i o r it i s e d g e n e s = 1x 1 0 -9 v ers us 0 .0 04 fo r oth er pr eviously implic at ed AD gene s), po si tive r e gula tion o f cy t ok in e produc ti on ( GO , 0 .009 v e r . 9 x 10 -4 ), c ellula r re spo ns e to cy tokine stimulu s ( GO , p =0.011 ver. 0 .0 0 9), cy t o kine -c ytokine re cep tor int erac t i o n (KEGG, p =1 x 10 -4 ve r . 7 x 1 0 -4 ), interle uk in-7 -media ted (GO , p=0. 053 ve r . 0 .048) a nd in terl euk in -4-m ediat ed s ig n alling pathw ays (NC I, p =0.01 1 ver . 0. 007) , inte r leuk in-2 ( Jen s en, p= 0 .003 v er. 0.035 ), in t er l eukin -12 (Je ns en, p =1 x 10 -5 ver. 0.0 01) and inte r l e ukin-23 compl ex (J en sen, p = 7 x 10 -6 ve r . 0.010). The ge n e s in the cytok ine pa t hw ays id en t i fi e d by ADGA P P inc lude IL 6 R , IL 22, I NPP5 D, IL2RA , IF N G , IL 18R1 , IL 18RAP , IL 1RL1 and IL 7R . Si gnalli ng inv olved in regula tion o f r e spo ns e t o int er f e ron γ ( GO , p =0.039 ver . 0.0 4 3), JAK1 -/JAK2 - ST A T 3-int erac ting gene s an d J A K - S TAT si gna lling p athway in ge neral (KEGG , p=4 x 10 -5 ver. 2 x 10 - 4 ),gene s dow n s t ream o f N F -κB -RelA tran scrip tion f ac to r ( T RRU S T, p =0.036 ve r . 0 .012) a lso ove r l appe d be t w e en th e t w o gene s e ts, as did reg ulati on o f T ce ll di ff eren tia tion (GO, p=0 .011 ve r . 0.00 7), s e l ectiv e ex pr e ssio n of c hemoki n e rece p t or s duri ng T -cell pola riza tion ( Wi kiP athway s , p=0. 036 ve r . 0 .004) a nd Th1, Th2 (KEG G, p=2 x 10 -4 ver. 6 x 1 0 -4 ) , T h 1 7 ce l l d if f e r ent ia t i on ( K E GG , p = 4 x 10 -7 v er . 7 x 1 0 -4 ). Besid e the p r e viou sly me ntioned c ytokin e - r el a ted ge n e s , we f ound oth er t a r g et s in d iffe r e nt immune pa t h ways : e.g . STAT 3, SOCS3 , ETS 1 , TRAF3, TRAF6, IRF1 . We di d not fin d enric hmen t of ge n e s in any s p e cific type of immunity – with a ll o f Th1 , Th2, Th17, T h22 repre se nte d and pr e viou sly sh ow n to play a rol e in ce rtain s ub se t s of AD pa tien t s , de s p i t e ove ral l par ticul ar . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint importanc e of Th2 and T h22 [36,179 ,180] . We di d no t se e muc h in th e w ay o f B cel l- s pec i fic ef fec t s, des pite s elec tive e xpan si on of ce rtai n B- c ell sub se ts in A D p ati ent s[181 ]. Gene s con c erned w it h e st a blis hment o f t he skin ba rr i e r were ma r g in ally enrich ed for in A DGAP P (due t o the p r i or i t i sati on of co rnifi ed env el ope gen e s , HR N R a nd RPTN ) , bu t le s s t ha n the pr evio u s ly repor ted A D gen es (G O , p=0 .045 ver. 8 x 10 -8 ). The s ec o nd w ay w e va lidated our r e sult s was to te st i f our ca ndi dat e s inte ract ed with ea ch oth er a nd wi t h t he gen e s w it h e st a bl is hed rol e s in AD pa thog en es is. W e u sed STRIN G [182] to visual i se t h e hig hest-c o nfid ence in ter action s (Fi gu re 5 ) amo ng the top 3 pr i or i t i sed gen e s at e a c h loc us f r om ADGAP P and o the r AD gen e s prev iou sly i mp lica t e d ( Additi onal File 9 : Suppl emen t ary Ta ble 3). N ot sur p ri singl y, the a n alysi s r evea led an ext ensive n etwork t hat i nclud e d 25 AD GAP P prioriti sed gene s, ce ntred on ke y immune regula tor s, s u ch as ST AT3, ST A T 6 , SO C S3 , IRF1 , TRAF6 . I t in c l ud e d d i r e c t bindi ng inter ac tion s be twee n t a r g ets pri oriti sed in the cur rent AD G WAS a nd out s i de o f it – betw een INP P5D and FCER1G a s w ell a s FC E R 1 A , IL 7R / ST A T 3 a nd TSL P, IL 7R and TSL PR, IF NG and IFN G R1 ; and SOCS3 and IF NGR1 . How eve r, w hen it c a me to the ge ne s dir ec t l y taki ng par t in e stabli s hing skin barrie r, 2 nd ranke d gen e at the ep id er m al diff eren tia tion locu s - RP TN , w a s th e onl y one shown to inter ac t with th e l at e corni fie d enve lope ge ne s.

Previ ou s a nn ota tion s o f A D G W AS l oci h a ve be en limi t e d in thei r a bility to i d enti fy lik ely causal gen e s [11 ]. He r e we pr ov ide a t h oroug h a nd ob jectiv e inve s tiga t i on o f the 25 Euro pean AD lo ci, by integ r a tin g all r e l eva nt avai la ble da ta th a t ca n be u sed t o prov ide ev id enc e f or ide ntify ing pot enti al ly ca us a l gen e s , and c ombin e thi s da ta in s uch a way as to pro duce a ranki ng for ev e r y gene at eac h loc us. W e int end th at t hi s thor ough a nal ysis and rank ing w ill prov ide a pr i or i t i s ed l is t o f A D ca ndidat e g ene s f or fu r th er e xpe riment a l work, a nd the de ta il ed pre s enta tion o f e videnc e f or ea ch gen e wil l enab le a pprop riat e and t ailor e d f o llow- up inves tigati on s to b e de s i gned . For 10 loci the t op ranke d gen e i s not t h e gene clos e st to the ind ex GW AS S NP . 8 l oci ha ve a s i ngle sta nd -out can di dat e c au sal ge ne ( s c ore > 50% of the t op 1 0 gene cumul ative sco re ) a nd 7 g ene s s c ore partic ula rly hi gh (>700 ) a nd/o r have a pa rticu lar stand - o ut s c o r e ( >75%). The s e ar e IL 6R at 1q 21.3, IL 18R1 / IL18RA P at 2q12 .1 , PR R5L at 11p 13, IL7R at 5p13.2 , PP2R3C a t 14q13.2 an d TR A F 3 at 14q3 2.32. Whil st in ma ny c a se s o u r a nal ysis s t reng t he n s t he evide nc e fo r ex is t i ng ca ndidate c au s al gen e s a t th e s e loci, for 6 loc i our sco re ra nks al te r na t i ve c andida t e s a s t he mo st lik ely c ausal g en e. One o f the s e 6 c an be con side red an in te re st i ng va lida tion o f our app roac h. IL1 5RA was pr evi ous l y co ns id e r e d the mo st pl au sible c andid at e gene at t he 10p 15.1 loc u s due to th e l imited e QTL evide nce that wa s a vail abl e a t t he tim e. Ou r ADGA PP a pp r o ach howev e r pr i or i ti sed IL 2 R A o ve r IL15RA. S i n ce the pub lica tio n o f th e GWAS in 20 15 thi s locus ha s be en f ollow ed up w ith CRIS P R expe r ime nts , whi ch report ed th at th e T-all el e a t r s 6 18 39660 down-regul a te s I L2RA ex pre ssion [ 144], s ugg e st in g that A D G A PP ’ s pri orit is ation at thi s loc u s is corre ct. Ot h e r valid ati on s of ou r appro ac h are p r ovid ed by t e st s o f enrichme nt o f ont olog y t e r ms a n d ev idenc e o f prot ein -pro tein in ter ac tion s a mongs t the top rank e d gene s a cro ss al l loci . Enric hment (among s t th e top 3 pri oriti sed gene s a t e a ch lo cus ) was found for the fol lowing o ntology term s : skin barrie r int egrity , T hel pe r cel l pola ris atio n, c ytokin e sign all ing an d JAK -STAT signa l ling. JA K- S TAT signal ling ha s r ec ently b e en c on firmed a s enric hed for among gen es p riori ti sed for inflamm ato r y s k in dise a se s (incl udi ng AD ) with H iCh IP -de rived T ce ll enha nce r c onnec tome [183 ] . Fur thermo re, ou r . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint enric hmen t ana ly sis highlig h ted t he rol e of JAK1, JA K 2 -int er a cti ng gene s, and in te re s ting ly the tw o gen e s the m s el ve s we re shown to con tai n an ov er-repr e sen tati on of rar e c oding varian ts in a n ew AD st u d y[184 ], w hich under s c or e s th e func ti onal import ance o f the p athw ay i n AD. Analo gou s l y, in inve stiga ting prote in -pro t e in int erac t i on s (u sing the STRI NG da tab a se) a mong ou r ADGAP P -prio riti se d c andida te gene s an d ot h er establ i s h e d can dida te s int erac tion s be tw een ge n e s wi t h immune reg ulati on (bu t not skin ba r r i er) fun c tion s were foun d amo ngs t the e st abli shed A D pla yer s : TSLP and i ts rec e ptor, TL R 2 , STAT6 , IL4 a nd inter fe ron γ rec ept or. Som e ro le s in skin mai nt e na nce ar e sup por t e d by s ome of our cand ida te ge ne n etwork modul e s. Th e gp130 -S O CS 3- ST A T 3 ax is pr om ot e s pr olif e r a tio n, migra tion and di ff ere ntia tion o f kera tinoc yt es [1 85] in skin -w ound hea ling, a ctiva t e d among st oth er by IL-6. S O CS 3 neg a tively regula t e s S TAT3 a ctiva tion by simult aneou sly binding t o pho sphoryla te d JAK k inas e an d gp130 cytok ine rec epto r [18 6 ]. ST RIN G d ata i s no t enti rely c ompreh en si ve and omi ts oth er func tio nal rela tion shi ps be tw een prioriti se d gene s. E TS1 a ctiva t i on r e s ul ts in d ownregul ati on of IL -6 pa thwa y and in ST A T 3 phosp horyla t i o n [ 1 87] . SO CS3 antago ni se s IL 1- depe nden t a ctiva t i on of NFkB pa th way by inhi biting ubiqu it i na tion o f TR A F 6 in th e TRAF -6/TA K1 c omplex[1 88]. LG P2’ s ne ga tive r e gula tion of t y pe 1 inter fe ron re sp on s e i s exer ted t hrough a ntagonizing TRAF ubi qui tin lig a s e ac tiv ity, inc luding TRAF3 and TRAF6[189 ]. Fin al ly, S TAT3 is a reg ul ator n ec e ssary f or ma int enanc e o f ST M N 3 e xpres s io n, and its ph o sphor y la tion a s w ell a s e xpre s s io n correla te s with S TMN 3 expre s s i on [190] . ST M N3 i nt e r a ct s wi t h STAT3 via i t s C -t erminal t ub ulin -a s soc iating r e gion [191, 192 ]. In g ener al, th e r e s ul t s of ou r GWAS prio ri t i sation analy sis remind u s tha t i nte r pr et ation o f a G WA S loc us i s c omplic at ed due t o varyi ng regul a tion be twe en ce ll type s an d wide s pr ead coregul atio n tha t mak es ide nti fic ati on of the t rue c au sal g e ne dif fic ul t. Ind ee d , recen t G WAS ex plo sion r evea l s t h a t on top of ea ch l ocu s being abl e t o con tain m ultiple sign al s[19 3] , eac h signal can infl u enc e multipl e co- regula te d gene s[194 ]. A s s o ci at i o ns w i t h m ole c u la r p he n o t y p es f o l lo w t he sa m e p a tt e r n , w i t h at le as t 9% o f h u m an eQ T Ls quan t i fi ed to con t a i n se conda r y s i gnal s [1 95] an d multipl e gene s impl icat ed for 5 0 % huma n e QTLs [14] . It i s w ell e s tabli she d tha t g en es in the same locu s show o ften co rrel at ed e xpre ssion, esp ec ially fo r gene s loc ali se d in t h e s a m e TAD[196 ], a nd tha t correl atio n it s el f i s h er i t a ble [197] . Acc ording to th e multipl e e nhanc e r varia nt hy pothe s i s , s eve r a l va r ia nt s in LD c an i nflue nc e mu lt i ple enha nce rs a nd c oope rative ly a ff ec t exp res sio n of targe t g ene ( s ) . Corradi n et a l . (2 014) provide ev idenc e f or it in 6 au toimmune d i s ea se s , inc luding RA, Cr ohn's di se a se and SLE [1 98], wh ile two adja cent thy roi d-rel a ted ge ne s a re r e gul ated by an enh anc er h aploty pe at a t hy ro id ca ncer ri sk loc us[19 9] . Ther ef ore, i t i s not surp r i s ing that many o f our loc i showed multi ple c o loca lis ation s for diff er ent ge n es and ti s s ue s, e spec ially in g ene-d en se r egio n s , w ith t he c avea t t ha t not all ma y be ca us a l . A r ece nt an aly s i s o f th e T W AS col oc ali sation me thod cl aim s th at ar ound 7 5 % of hit s wil l be non-c au sa l in the in stan ce o f c orre lat ed ge ne ex pr e ssion a t th e locu s[145 ], a nd w e hypothe size tha t may be the ca se a t lo ci 11 q13.1, 1 4q13.2 , and 20q1 3.33, w her e ex pr e ssion o f a s ma ny as 4-6 gen e s co local ise s w ith AD G W AS sign a l in the T WAS re s u l ts, a lon e. S t i ll, due t o a di stinc t pos s i bili ty of detec t i on o f multipl e targe t ge ne s a nd va r i ant s a t a loc u s , we do no t focu s only o n top- r a te d hit s in our ge ne and vari ant r ank ing. A D GW AS l oc i whi ch w e beli eve shoul d be fu rth er e xperime nt a lly inv es t ig ate d in tha t reg ar d, inc lude : 2q 12.1 ( IL 18R1 , IL 18R AP, IL 1R1 ), 5q 31.1 ( KIF 3 A , PD L I M 4 , SL C22A4 , IR F 1 ) and 20q13.3 3 ( STM N3, LIM E 1 , ARFRP 1 ) – t he fi r st t w o es pecia lly d ue to co nta ining at lea s t two ind epend e nt s ig nal s in th e G W AS an alysi s . Mos t o f t he gen e s w hich s how e QTL c olo ca lisa tion ac r oss many ti ssu e s indic at e th e sam e direc tion of e ff ec t, s uc h a s in t h e ca se o f P RR5L , a top prio riti s e d hi t at the 11 p13 loc u s , wh ere th e pr o tec tive all ele i s a s socia ted with in cr ea sed ex pr e ssion in the s k in, w hole blo o d and i mmune c ell sub s et s. Howev er, th e re sul t s a t thre e loci ( 2 q12. 1, 1 4q13.2, and 20q1 3.33 ) imp ly ther e ma y be tis su e - . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint depe nd ent e f fect s on ex pre ssion. Pr o tec tive a llele s in t h e s k in e QTL s showe d po si t i ve ef fec t and neg ative e ffe ct on ex pre ssion i n blood e Q TL s in STMN3 a n d LIME1 , whi le reve rs e patt ern wa s s een for AR F R P 1, and simila r ly , ch ange o f s ig n betwee n ti ssue typ es w a s al so se en f or IL 18RAP and PP2R3 C . This indi ca te s t hat ca u s a l va r i a n t s p oten tia lly re sid e i n ti ssue type - s pec ifi c r eg ulatory region s a nd cont ex t-de pe nden t e ff ec t o f the se gen e s c ould im pact a topi c derma ti tis ph eno ty pe. Howev er, eve n when focu sing on a singl e t i s s ue o f in t e re s t, we fre que ntly ob serv e lack of co ncordanc e in re sult s among s ame tissu es from di ff ere nt da ta se t s , whic h was pa r tic ularly clea r i n the e QTL colo cali s a tion analy s i s, w her e we us ed e Q T Ls for t h e s a me ti s s u e s so urc ed from di f f ere nt st u d ie s : s k in and LCL f r om bot h G T Ex a nd Tw ins UK, w hole bl ood from GTEx and e QTL Gen, C D4+ T ce lls fr om Blu eprin t and CE DAR. H i ts suc h a s SL C 2 2A 5, PRR5L , IL2 RA , TRAF3 , F AM 177A1 showed strong coloc ali sa tion in one d a ta se t w hic h was th en mi ssing i n ano the r. Thi s c ould be expl ained b y a hos t of fact or s r e l ating to s tudy d e s ig n, e x perimental m e t h od s a nd d ata a naly s i s c ho ices . The fi r s t two i nclude sa mple size a nd c on sequ entl y s ta tis t i cal powe r, de mogr aphic comp o sition o f studie d co hor t, di f f e re nce s in ha r v e s t e d cel l com po s i tion and c ondi tion, an d tech nic al d et ails d ete rmining sen s i t i vity of the a ssay . I n the third c a teg ory, choic e s r e gar ding data c le an- up suc h as quali ty contro l filt er s and da t a analy s i s pa ra met ers such a s con found er s and g enom e w indow s i ze us ed fo r e QTL detec tion ca n al so pr eve nt S NP inc lu s i on in QTL analy s e s and th u s a ffe c t our abil i ty to inter rogat e the da t a f or a par ticu lar as soc iation . N ev er th el e ss, c on s i st ent e QTL coloc alis ation repl icati on acro s s multipl e da ta se t s a nd ti ssue s , w heneve r se en - a s i s t h e ca se e .g. f or IL 6R in w hole blo od ins t i l s u s wi t h mor e confi d ence t ha t t he e f fect s ee n i s re a l. Our analy s i s c onfi r m ed many loc i whe r e t he re is pl eiot ropy be t w een a t op y and in flammat ory dise a se s. Fo r thr ee ge n e s , th e direc tio n o f e ff ect ma t c h es : DEXI (T1 D, MS ), SOCS3 (IB D) , ET S1 (S LE), but fo r many othe r s the di rectio n o f e f fe ct is oppo si te . Thi s i s no t su rpri s in g, a s d espi te ge n erally oppo sing t y pe s of immun e activ atio n in p s o ria si s ver su s e czema (Th1 7- s ke wed v er s u s Th2- skew ed ), ye t 81% of d if f e r e n t i ally expre s s e d gen e s in A D a r e als o DEG s in p s o ria si s [200] . W e obs erve s uc h co nt ra s ts for s eve r a l immune re gulat or s : I N PP5 D (IB D ) , TR A F 3 ( IBD ), IL 6R ( RA ), IL7 R ( M S), IL2 RA (T1D) , CL EC16A (MS) whic h prob ably ste ms fr om dif fer ent Th1 /Th2 / Th1 7 bala nc e i n adaptiv e immune s y s tem dy sr eg ul ation i n tho s e di s e ase s[201– 204] . A s t reng t h of ou r varia n t pr i or i ti sa tion a n alysi s is tha t we used th r ee di f fer ent B aye s i a n varia n t fin e- mapp ing method s: JAM, Pai n t or a n d Fin e map. W e in t e gra te d the thr ee met hod s (ran w ith di f fer ent parame t e rs ) in li ght o f previ ou s ly sh own inc ongruenc e in the ir re s ult s r e s ul t i ng f r om diff eren t model as s u mp t i on s[205 ]. A not able cav ea t in o ur fine -mappi ng of varia n ts is ex clusion o f i ndel s and bi gger struc t u ral varia n ts. The re fore , variant p ri oriti sa tion r es ult s d o no t aim to p rec is ely i denti fy the c au sal va r i ant s bu t fi nema p the b r oa d er r e gion w ith the ai m o f priori ti s in g SN Ps whic h in turn ma y provi de addi t i on al ev idenc e for wha t gene s in th e r e gion a re likel y to be involv ed . Our fin e-mappi n g wa s a l s o re st ric t e d by th e power an d impu tati on pa nel (1,0 00 Genom e s) o f th e publ i s he d EAGLE GWAS [11 ]. Future GW AS w it h bigge r s amp le siz e an d den ser va r i an t imput ation s wi ll inc re ase the pow e r of futu re f ine -mapp ing an d, in turn, c au sa l ge ne iden tific a tion a t t empt s. Our ADG AP P pipe lin e fo r foll o w-up o f G WAS s ignal s, w hils t fo cus ed on th e int egr ation o f A D- relev an t re s ou rc e s in thi s u s e ca se , c an b e ea sily adap ted f or ot her di s ea se s or tr a its, fol l owing iden t i fi cati on o f t he mo s t r ele van t a nd p owerful mol ecu lar da ta se t s . F o r some tra i t s (e .g . blood - relat ed) the re are very well-p o wered a n d informa tive acc e ssible da ta se t s ( e.g. e QTLG en) w hile the more s pec ific da t a may still be la ckin g, fo r ins t a nce di ff ere nt laye r s and cell t y pe s within t he s k in. Bet ter pr edic t i o n s ma y be made in th e fu tur e by i nte grati ng ev idenc e f rom ac r o s s more ti ssue ty pe s , esp ec ially at a singl e -cell re s olu tio n – suc h da t a s e t s are a lr e ady be ing g ener ate d f or rela ted diso rder s , suc h a s a s t hma [206] ; c ons i de ri ng tran s- a nd is of or m - l evel mecha ni s m of actio n , and ex plici t l y model ling n etwork c onn ectiv ity vi a protein -pro tein in te rac tion s a nd c o- e xpres s ion . . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint Whils t th er e a re lim ita t i on s in ou r appro ac h, as o utlin e d in the s ec t i on s a bove , we ne vert hele s s fi nd A D G A PP us ef u l t o e as i l y f la g t he ge n es wh er e w e f i n d m os t ev i d e n c e wh i c h ca n t he n b e ca r ef u l l y ev aluat ed. Fur t h e r mo r e , us i ng our s t a nd ardi sed scoring sys tem to as s e ss priori ti sa t i on, we can u s e a b s ol u t e s c or e t o m a ke d i r ec t co m pa r is on o f s t re n g t h of p r ior i t is at i o n ac r os s and w i t h i n l o c i , % o f cu mulative s c o re ca p t u r ed by t op g ene f ac ilitat e s com pari s on o f s tr ength o f e vid enc e fo r ge ne s wi t hi n the loc u s . W e u s e a c ombin ation o f t h e two u se ful me tric s to prop o se th e mo s t pl a u s ibl e ca ndidat e g ene s in th e A D GW AS. L oci w here w e ar e more c on fid ent i n pri or i t i sat ion of s in gle g en es , lend them se lves to dir ec t experim en t a l i nve s tig a t i on , such a s TR A F3 at th e 14 q 3 2.3 2 l o c u s a nd PRR5L at t he 1 1p13 loc u s . Addi tiona ll y, inv es tig ating l oci w ith cl ea r ca ndidat e g e nes a nd a s socia t i o n wi t h multiple i n flammato ry di sea s e s sho wing consis ten t direc tion o f e ff ect, such as 11p13 ( PRR5 L – MS, a s thma ), 11q24 ( ET S 1 - p soria si s , cel i ac dis e a s e ) and 16p 13.13 ( DEXI and CL EC16A - T1 D , M S, alo pecia a rea ta, S LE, a st hma ) may r e veal pr omi s in g t arge ts wit h pot entia l d rug re purpo sing fu ture . Ot h e rs wi t h oppo sing dir ec tion s o f e ff ect s may r e veal po ten tial a dve r s e s id e e f fec ts for co ns id e r a t i on in th erap eu t i c d ev elopme nt (e .g. wi t h an ti- IL6 biologi cs for R A ) .

Through our A D GA P P pi pelin e, we ha ve de velope d a n i nt e gr at ed appro ac h wh ich ha s a llow e d se mi- quan t i fi cati on o f t he s ubj ectiv e p roce s s o f priori t i s a tion o f gene s ( and varian t s) a t AD G W AS loci. W e hav e ama ss ed 103 data s et s (i n cludin g molec ular da t a s e t s f r o m r e leva n t t i s su es) a nd use d the s e to gen era te pri oriti satio n s c or e s fo r ea ch g e ne a t th e 25 e sta blis hed E ur op e an A D G W AS loci . A t eight loc i we were a ble t o priori ti se single can di date ge n e s w ith good evide nc e, many n ot being t h e cl os e s t gen e to th e GWAS index SNP . F or 6 loci we i dentify g ene s no t previou sly i mplic ated a s th e top- ranked can di dat e s . We pre se nt com prehen siv e r e s ul ts an d dis c u s s ion o f th e evi dence at eac h loc us to enabl e appr opria te follow - up mole cular bi ol ogy investiga tion s, in cludin g drug targe t discov ery o f th es e pot enti a l ca u s al gene s . Th e princi ple s be hind t h e ADGA PP pi p el ine c an b e adop ted in ann o t a t i ng GW AS on othe r di se as e in t h e fu tur e . Additional File 1: Supplementary Text. D escri ption o f A DG A P P s c oring sys tem. Additional File 2: Supplementary Figure 1. Ove r v iew o f con tribu tion o f di ffe r en t t yp e of - omic s ev idenc e to gen e and S NP r a n king in all th e AD G WAS loci . Each bar r epre se nts a n umber o f indi vidual studie s in a gi ven c at egory contributi ng to th e s um tota l of evid ence. Additional File 3: Supplementary Figure 2. Di str i bution o f scor es obtain ed in our model fo r ca ndidat e g ene ra nk ing a t eac h locu s (l a b elled w ith le a d SN P) . Additional File 4: Supplementary Figure 3. Di str i bution o f scor es obtain ed in our model fo r caus al SNP ra nk ing a t each locu s (l ab ell ed w ith l ea d SN P) . Additional File 5: Supplementary Table 1. C oloca li s a t i on r e s ul ts from co loc a nd T WAS me t h od s (eit her ba se T W AS or TW AS -ba sed co loc ) on A D G WAS an d e Q T Ls in ti s s u e s from a number o f data se t s . We rep ort coloc r e s ul ts for gen e s w ith p os t e r io r p roba bili ty of a shar ed ca us a l e QTL a nd GWA S vari a nt ( P PH 4 ) > 0 . 5 for a ny g ene w it h at lea s t one st rong c oloca li s a t i on r e s ul t of PPH4 > 0 .9 , whi le for TW AS w e sh ow g ene s with g en ome-wi de signi fic ant an d indep ende n t coloc ali satio n ev idenc e in c onditi onal and jo in t analy s is . Gen e r a nk at the given locu s in our final GWAS ge n e p r i o r i t is at i o n m o de l is a ls o g i ve n . Additional File 6: Supplementary Table 2. Re s ult s mee ting th e ge nome -wide s ig n ifica nce thre shol d wi t h no signi fic ant h et erogen e ity in HEI D I a nalysi s for SM R te s t of G T Ex, Bl ueprin t and e Q TL Gen . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint e QTL ins t rument s aga i nst A D GW A S . Gen e rank at t he gi ven l ocu s in our fi na l G WAS gene prioriti sa tion mod el i s al so giv en. Additional File 7: Supplementary Text 2. ADG A P P prio riti s a tion s umma r y fo r AD l oci. Additional File 8: Supplementary Figure 4. Score by type s of e videnc e in all th e A D GW AS loc i. Sc ore s f o r top 3 ra nked gene s a t ea ch lo c us ar e s hown p artition e d by c atego ry of evid ence – he re inc luding t h e top 10 c a tegori e s c ontr i b uti ng the hi ghe st propo rtion o f t o ta l s c or e at th e top 1 0 ranke d gene s for al l loc i . Or d e r o f loci cor r e s pond s t o the o rde r i n Tabl e 1. Additional File 9. Supplementary Table 3. G en e s p rev iou sly imp lica t e d in AD, n ot prioriti s ed by ADGAP P. Additional File 10: Supplementary Table 4. Compari son o f Enric hr-b a sed s ig nifica nt enrichme n t t es t in g res ul t s fo r : t o p 3 ge n es c o mb i n ed f r o m a cr os s a l l t h e AD G WA S lo c i t est e d ( G en es .t o p_ 3 ) a n d other kno wn A D gen e s (G ene s.previ ou s ly _know n), aga in st s ele c t ontol ogi e s . L i st o f abbrevi at i on s AD: a topic de rmati ti s; A D G AP P: A topic D ermati ti s GW AS Anno ta tion P r i or i t i sati on P ipeline ; ca QTL - ch r oma t i n ac ce s s i bili ty qua nti t a tive trai t loc us; AS: ank ylo sing s p ondy liti s; CD : Cro h n’s di sea s e ; D GE : diff er ential gen e expre s s io n ; e QTL: e xpre ssion qua ntit ative t rai t l ocus ; GTEx: G e n o t y pe-Ti s s ue Ex pr e ssion pr ojec t ; GW A S: genome -w ide a ssocia t i on s tudy ; h QTL: hi stone q u anti t ative tr ait lo cu s; LC L – lymphobl astoid cell line ; L D: link ag e d iseq uilibrium ; MHC : maj or hi s toc omp atibili t y comp lex; mQTL: DNA met hyl ation qu an t i tativ e trai t loc us; MS : multi ple s c lero s i s ; P P: po s t er ior probabi lity; pQTL: pr o tei n qu anti ta tive tr a i t locu s; RA : rheuma t oi d ar thr i t i s; S LE: S ys t e mic l upu s er y thema to su s ; SNP: singl e nuc l eoti de pol ymorphi sm s ; s QTL: splic ing q uanti ta tive t rai t l ocu s; T1D : type I diab ete s; UC: ulc e rative c oli ti s Declar at ions Ethic s a ppro val a nd c on se n t to par t i cipa te Not applic able Consen t for pub lica tio n Not applic able Avai la bility o f da ta a nd m ateri al s All th e c ode w r i t t en t o ca rry out th e ADG AP P an a lysi s is avai lable on GitHub at: h ttp s: //github .c om/ M RC IE U /ecz ema_ gwa s _ fu Code i s archiv ed u nder : ht tp s : //doi .org /1 0.5281 / ze nodo.37 75865 The data se t s suppor ting th e c onc lus i o n s of t hi s a rtic le a re a vail abl e i n th e Fig s h a re r e po s i tory. Dataset S1. R ef er enc e ta ble de tailin g ind iv idual data se ts u sed in A D G AP P. http s : / /doi.org/10 . 6084/m9.fig s ha re.122 22731 Dataset S2. The final e videnc e da ta se t us ed in r a nking gene s an d vari an ts: http s : / /doi.org/10 . 6084/m9.fig s ha re.121 30701 Dataset S3. L ocu s Zoom- s tyle g ene a n d varia nt s c o r e plo ts f o r e ach locu s: Gene : ht tp s: / /d oi.org /10.608 4 / m9 . f i gsh a r e .122210 06 . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint Vari ant: h ttp s ://doi .org/10 .608 4/ m9 .fig sh are.12221 033 Dataset S4. Ful l gen e and v arian t loc i ran ki ngs: Gene : ht tp s: / /d oi.org /10.608 4 / m9 . f i gsh a r e .121308 24 Vari ant: h ttp s ://doi .org/10 .608 4/ m9 .fig sh are.12130 857 Dataset S5. Da ta e videnc e for top 10 -ran ke d gene and v arian t s at eac h loc u s : Gene : ht tp s: / /d oi.org /10.608 4 / m9 . f i gsh a r e .121308 63 Vari ant: h ttp s ://doi .org/10 .608 4/ m9 .fig sh are.12130 878 Comp e tin g inter est s TRG rec eive s funding f r o m G l axoS mithKli ne and Biog en for un rela ted re sea rc h. F u nd in g This w or k wa s suppo rt ed by a Spring boar d aw ard (SBF0 03 \ 1 094), aw a rded to L P, s upporte d by the A c ad e m y of S c ie n ce s ( A MS ) , t h e We l lc om e T r us t , t he G o v er n m en t D epa r t men t of B us in es s , E ner gy and Indu strial St rat egy (BE IS ) and t h e Bri t i s h H ea rt F ounda tion (BHF ). T RG ack no wledg es s upp ort from th e UK Medic al R e s ea r c h Counc il ( MC_U U _ 00011 /4). T GR is a U KRI Inn ova tion r e s ea r c h f ello w (MR / S 003886 / 1 ). T winsUK i s fu nded by t he Wellc ome Tru s t , Me dica l Re se arch C o uncil , European Union, Ch roni c Di s ea se R es earch Fo unda tion (C DRF ), Zoe Globa l Ltd a nd th e Na tio nal In s ti tut e fo r Health R e s ea r c h (NI H R) - fund ed BioR es ou r c e, Cl inica l R e sea rch F ac ility a nd Biom ed ic al Resea r c h Cen tre ba sed at G uy ’ s and S t T homa s’ N H S Founda t i on Tru s t i n par tne rship with King ’s Coll ege Lo ndon. Aut hor s’ contri bu tio ns LP initiat ed t h e study , LP, MKS and TR G contribu ted to th e de s i gn of the stu dy. M KS and T G R perf ormed the a n alysi s o f the d a t a . MKS w r ote t h e manu scrip t wi t h input from LP and TRG. All author s re ad an d approved t h e manu scri pt be fore submi s sion . Ack nowle dge me n t s We ac know ledge th e many open a cc e ss data se t s t ha t have made thi s wo r k po s s i ble . W e al so ac know ledge e Q T LGen for p rovidi ng e QT L r esul t s be fore thei r publ icati on and Tw insU K ( ht tp: / /ww w.twin suk. ac .uk / d a t a - a cc e ss /p ublica t i on s/ ) for pr oviding ac ces s t o dat a for e QTL ana ly s i s. Fi ne-map ping anal y s i s u s ed th e U K B ioba nk genetic dat a re sou rce a s an LD r ef ere nc e [Applic ation #10 074].

1. Hay RJ, Jo hn s N E , Willia m s H C, Bol lige r IW, D ell aval le R P, Ma rgo li s D J , e t a l. Th e globa l burden o f skin di s ea se i n 2010: An anal y s i s of the p r e vale nce a nd impa ct o f skin c ondi tion s. J Inv e s t Derma tol. Elsev ier Ma sson S AS ; 2014 ;134:1527 –34 . . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint 2. K im JP, Ch a o LX, Simpson EL , Sil verb er g JI. P ersi ste nce o f ato pic derma titi s ( AD ) : A sy ste matic revie w an d meta -a nal ysi s. J Am Acad Der matol. 2016;75 : 681-687.e 11. 3. F urue M . T h elper type 2 s i gna tur e s in atopic de rmati tis . J Cuta n Im munol Alle rgy . 2018 ;1:93–9. 4. Gho sh D , Ber ns t e in JA, Khur ana H e rsh ey G K, Rot h enberg ME, Me rsha T B . Lev er aging Multilay er ed “Omic s ” Dat a fo r Ato pic D e r m a t i tis : A Ro a d Map to Prec i s i on Medic ine . Fron t Im mu nol. 2018 ;9:2727. 5. Bi cke rs DR , Li m HW , M a r g oli s D , Wein stoc k MA, Go odma n C , Fa ulkne r E , et al . The burden o f s k in dise a se s: 2004. A joint projec t of th e Am erica n Acad e my of D erma tolog y As soc iat ion a nd the S oc iety for Inve stiga t i ve Derm atol ogy . J Am Ac ad Derma tol. 20 06;55 :490 –500. 6. Druck e r AM, W ang AR, Li W Q, Sev e t s o n E , Bloc k JK, Qu re s hi AA . The Bu rden o f At o pic D erma titi s: Summ ary of a Rep ort f o r the Na t io n al Ec z e ma A ssoc iati on. J Inv e st Derma tol. Th e Author s; 2017 ;137:26– 30. 7. Brunne r P M, Silv erbe rg JI, Gut t m an -Ya s s k y E, Palle r AS , Kaba shima K, Amag ai M , et a l. Inc r e as i ng co mor b iditi es s ug ge s t t ha t ato pic derma titi s i s a s y s t emic dis orde r. J Inve st D erm atol. 20 17;137 :18 – 25. 8. Wu L -C , Hwang C-Y , Chu ng P -I , Hua T- C , Chen Y -D , Chu S -Y, et al . Au toimmu ne d ise as e co mor b iditi es in p atie nts w it h a topic der matiti s : A n ationwi d e ca se -c ont r ol study in Ta iwan. Pedi atr Allergy Immunol . 2014;2 5 :586–92 . 9. S chmitt J , S chw arz K, Baur echt H, Ho tze M, Föl st er - Hol st R , Rodríg ue z E , et al . A t opi c de r ma t iti s i s a ss oc i at e d w i t h a n i n c r e ase d r is k f or r h eu m at o id a rt h r it is a nd i nf l a m m at or y b o w e l d i se as e , a n d a dec r e a se d ri s k fo r t y pe 1 diabe te s . J All er gy Clin I mmunol. 20 16;137:1 30–6 . 10. Elmose C, T hom s e n SF . Twin Studie s of A topic D e r ma t i ti s: In te rpre ta tion s and Appl ication s in the Fi laggrin Era. J All ergy . Hindaw i; 2015 ;90 2359:1 –7. 11. Pat ern oste r L , Standl M, Waage J , Ba urech t H, Ho tze M, Str ach an D P, e t a l. M ulti- anc e s t ry gen ome-wid e a ssoc iati on s tudy of 2 1,00 0 c ase s and 95, 000 c on t rol s ide nt i f i e s ne w risk loc i for a topic dermati ti s. N at Gene t. 2015 ;47 :1449– 56. 12. H ormozdia ri F , Ga z a l S, Va n D e Geijn B, Finuc ane HK, Ju CJ T , Loh PR, e t al . Lev eraging molec ular quan t i tativ e trai t loc i to und er st a nd the ge netic a r c hi tec ture o f di s e a se s an d com pl ex trait s . N a t Gene t. 20 18;50 :104 1–7. 13. Ricañ o - P onc e I, Zhern ak ova D V. , D e e len P , Luo O , L i X, I s aa c s A , et al . Re fined m apping of autoimm une di s ea se a s socia ted gen etic varia nts w ith gene expre s s io n s ug ge sts a n importan t role for non-c oding R NA s. J Auto immun. 20 16;68 :62– 74. 14. G a maz o n ER, Se grè A V, v an de Bu nt M , W e n X, Xi H S , Hor m oz di a r i F, e t a l. U s i ng an a t l a s of ge ne regula ti on ac r o s s 44 huma n ti ssue s to inf orm c omplex dise ase - and tr ait -a s s oc ia te d v ariation . Na t Gene t. 20 18;50 :956 –67. 15. G rundb erg E, Sma ll KS , Hedman ÅK, N ic a AC , B uil A, Ke ild s o n S , e t a l . Mappi n g ci s -an d t ran s - regula to r y ef fec t s ac r o s s multipl e ti s sue s i n twins. Nat Gen et. 201 2 ;44:1084 –9 . 16. M o r ri s A P. Tra ns eth nic meta -an a lysi s of genomew ide a ssoci a t i on studi e s. G en e t Epidemi ol . Wiley O nl ine L ibr ary; 201 1;35:809 – 22. 17. Boyle A P , H o ng EL , Harihar an M, C he ng Y, Sch aub MA, Kas owski M, e t al. Ann ot a ti on of func t i o nal v aria tion in per sonal genome s using Regulo me D B . G e nome Re s . Cold S pring Harbor L ab ; 2012 ;22:1790 –7. . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint 18. W a r d LD , Kell is M. H a plo Reg v 4: Sy s t ema t i c min ing of pu ta t iv e c au sa l va r ia nt s, c ell t y pe s , regula to rs and ta r g e t gene s f or human c omple x trait s an d dis ea se . Nuc leic Acid s Res . 2016 ;44: D877– 81. 19. Võ s a U , C lari ngboul d A, W e s t ra H- J, B onder M J, D eel e n P, Z eng B, e t al. U nrav el ing the p olyge nic arch itec tur e of c omplex trai ts u sing b loo d eQTL met aanal y si s . bio Rxiv . 2018; 447367. 20. G TE x Cons orti um. G enet ic ef fec ts o n ge ne ex pr e ssion a c ro ss huma n ti ssue s . Na ture. 2017 ;550:204 –13. 21. Chen L , G e B, Ca sale F P, Va s que z L , Kwa n T, Garrido -M artí n D, e t al. G e n etic D river s of Epi gen etic and Tran s c ripti onal Va r i a tion in Huma n I mmune C ell s. Cell . 2016;16 7:1398 -141 4. e24. 22. Benne r C, S p ence r C C A , H a vulinn a AS , S alomaa V , Ripa tti S , Pirine n M . FI NEMA P: Ef fic ien t va r i able sel ec tion u sing summary data fr om ge nome-wid e as soc iation s tudie s . Bi oinfo rmatic s . 2016 ;32:1493 –501. 23. N ew combe P J, Conti D V ., Ri char ds on S. J AM : A S cal abl e Ba ye sian F ra mework fo r Joi n t A na ly si s of Margi n al SN P E ffe ct s. G en e t Epid emio l . 2016; 40:188–2 01. 24. Kicha ev G, Roy t m an M , John so n R, E ski n E, L indst röm S , Kra f t P, e t a l. Improve d method s f or multi- trai t fin e m apping of pl eiot ropic ri sk loci . Bioinfo rmatic s . 2 017;33 :24 8–55. 25. G u sev A, Ko A, S hi H, Bhati a G , C hung W, Pe nni nx BWJH, e t al . Int egrative a ppr oac he s fo r l arge - scal e tr a n s c r i pt om e -wide a ssoci ation s t u di es. N a t G ene t . Natur e P ubli shi ng Grou p; 201 6;48:245– 52. 26. Richa r d son TG , Hema ni G , Gau nt TR, Re lton CL, D a vey Smith G. A tran scrip to me-wid e M endeli a n randomiza tion study to uncov er t i ssu e-d epend e nt re gulato ry me chani s ms ac r o s s the huma n pheno me . Na t Commun. 2020;11 :1 85. 27. Pivido r i M , R ajag opal PS , Barb eira A, Lian g Y, Me lia O, Ba s tarach e L, e t al . Ph e nomeXca n: Mappin g the ge n ome to the ph enome t h rough the t ran sc r ip tome . bioRxi v. Col d S pring H a r bor La borat or y ; 2019 ; 28. Li MJ, S ham P C, W a ng J. G ene tic vari a nt r e p r e se nta tion, a nnot ation a nd prior itiz a t i o n in the p o st - GWA S era . Ce ll R e s. 2012 ;22:1505 – 8. 29. W a ta nabe K, Ta ske se n E, Van B oc hoven A , P osthuma D. Func ti onal ma pping a nd anno t a t io n o f gen etic a ssocia t i on s w ith FUM A. N at Co mmun. S pr i nger US ; 201 7;8:1–10 . 30. A m lie- Wol f A, Ta ng M, Mly na rs k i EE, K uksa P P, V a llada re s O , Katani c Z, et a l. I n f e rNo: Inf erri ng the mol ecula r mecha ni s m s of n onc oding genetic va r ia n ts. N ucl eic Acid s Re s. O x for d U n iversi t y Pr e ss; 2018 ;46:8740 –53. 31. Fac hal L , A s c hard H, B ee sley J , Barn e s DR, Alle n J , Kar S, e t al . F ine -mappin g o f 150 br e a s t c anc e r risk r egion s ide nti fi es 191 li kely t a r g et ge ne s. Na t G ene t. 20 20; 32. Liu B, G l oudem an s MJ, R ao A S , Inge l sson E, Mon t g omery S B . Abund ant a s soci a t io n s w ith gen e ex pr e s si on c omplic at e GWAS foll ow - up . N a t G en e t . Sp ringe r US ; 2019;50. 33. Shoo sht ari P, Hu ang H , Co t s ap a s C . In tegra tive G e netic a n d Epig enetic A nalysi s U n cove r s Reg ulato ry Mec hani sm s o f A u t oi mmun e Di sea se . Am J Hum G ene t. E l s e vier Comp any .; 201 7;101:75– 86. 34. Taşan M, Mu s s o G, Hao T, Vid a l M, M ac r a e C A, Ro th FP . Sele c ting cau sal g ene s from genome - wi de as s o cia tion s t udie s vi a func tio nally coheren t subne two r k s . N at Me thod s . 2015 ;12:154– 9. 35. Pell erin L, Henry J , Hs u C -Y, Ba lica S, J ea n- D e c os t e r C , M échin M - C , et al . D e fe c ts o f f i lagg r in - l ike . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint prote in s in both l e s i o nal an d nonl e sional atopic s k in. J All ergy Clin Immunol . 2 013; 131:1 094–102 . 36. Leung DYM, Gu t tma n -Ya ssky E. Dec ip hering t h e comp l exitie s o f a t o p ic de r ma titis : S hif t i ng paradi gm s in tre atme nt app roac he s . J All ergy Clin I mmunol. 20 14;134 :76 9–79. 37. Suárez -F ariña s M, Ti n t l e SJ , S heme r A , Chiric oz zi A, Nogr ale s K, Ca rdinal e I, e t a l . Nonle si onal atopic de rmati tis s k in i s c harac teri zed by br o ad t erminal di ff ere ntia tion d ef ec t s a n d va r ia ble imm une abno r ma li tie s. J Alle rgy Clin Immu nol. 20 1 1;127:954 - 9 64. e4. 38. Scha id DJ , Ch en W, L ar s o n N B. F rom gen ome-wid e a ssoc iati on s to ca ndid ate c aus al va r i ant s b y sta t i s t ic al fin e-ma pping. Na t Re v G ene t. Spring er U S ; 2018; 19:491–5 04. 39. Segrè A V. , G ro op L, Mooth a VK , D a ly MJ, Alt sh uler D. C ommon Inhe rit ed V ari ation in Mitocho nd r i al G ene s I s Not E nr i ch ed f o r As s o cia tion s w it h Type 2 Di abet e s or R el a t e d G l yce mic Trait s. Vi ss c h er P M, edi tor . PL oS G ene t. 2 010;6:e1 001058. 40. A u t on A, Ab eca s is GR, Al t s hu le r DM, D urbin RM , Ben tley D R , Chakr av art i A, et al. A globa l ref ere nce fo r human g ene tic varia tion. Natur e. 2 015;526: 68 –74. 41. Bycrof t C, F re eman C, P etkova D, Ban d G, El liot t LT, Sharp K, et al . T he UK Biob an k re sour c e w it h dee p phe notypi ng and genomic da ta. N a ture . Springe r US; 2 018;562:2 03 –9. 42. M i t c hell R , Hemani G , Duddi ng T, P at erno s t er L . U K Bioba nk Gen etic Data : MR C-I EU Q ual i ty Cont rol, Ver sio n 1. 2 017; 43. Kim SA, Cho CS , Kim SR, Bull SB, Yoo YJ. A new ha plo type block de tecti on met hod for d en se gen ome s eque nc ing da t a ba s ed on i nte r v al grap h modeli ng o f c luster s o f high ly c or rel a ted S NP s. Bioin f o rma t ic s. 2018;34 :388–97 . 44. A h Kim S, Bro ssard M, R o shand el D , P ater son A D , Bu ll SB, Yoo YJ . g part: huma n g enome parti tioning and v isua liza tion o f hig h-d en sity SN P da ta by id enti fying haplo type bl ock s . Bioin f o rma t ic s. 2019; 45. D a da ev T, Sa under s EJ , N e wco mbe PJ , Anok ian E, Leong amorn ler t DA, Brook MN, e t al. Fin e- mapp ing of pro s tat e ca ncer su s c ep tibili t y loci in a large me t a -an aly s i s ide n t i fi e s c a ndid ate ca u sal va r ia nt s . Nat Commu n. 2018; 9 :2256. 46. Slowi kowski K, Hu X, Ra ycha udhuri S. SNP s ea : An alg ori t hm to id enti fy cell t y p es , ti ssue s and pathw ays af fec t e d by ri sk lo ci. Bi oin form atics . 2014 ;30:2496 – 7. 47. Su AI, Wil t shir e T, Batal ov S, L app H, Ching KA, Bloc k D, e t a l. A ge ne a tla s o f t he m ouse and huma n prote in -enco ding t ran scrip tome s. Proc Na tl Ac ad S ci. 2 004;101 :6 062–7. 48. H en g TS P, P aint er MW , El pek K, Luka cs-Ko rnek V , Ma u ermann N , Turley SJ, e t a l. The Immunol ogic al Gen ome P r o ject: n etw or k s o f gene e xpr es s i on in immune cel l s . Na t Im munol. 2008 ;9:1091– 4. 49. Kawa ji H, Li z io M, Itoh M, Kan a mor i -K atayam a M, Kai ho A , Ni s hi yori -Suek i H, et al . Compar is on of CAGE an d RN A - seq tran s c r iptom e pro fili ng us i ng cl onally ampli fi ed and s ing le -mo lec ule ne xt - gen era tion sequ enc ing. G enome R e s . 20 14;24 :708–17. 50. de Le euw C A, Mooij JM , He ske s T, Po sth uma D. MAG MA: G e n eraliz ed G ene -S e t A na ly si s o f GWA S Dat a. PL oS Comp u t B iol. 201 5;11 : 1–19 . 51. Buil A, Br own AA , Lap pal aine n T, Viñ u ela A, Dav ie s MN, Zheng H -F , e t al. G e n e-gen e a nd ge n e- env ironmen t inte rac t i o ns d e te cted by t ran s c r i p t ome se quence analy s i s in t w ins . N a t G e ne t. 2015 ;47:88–9 1. . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint 52. A u lchen ko YS , Ripke S, I saac s A, van D ui jn CM. Gen ABEL : an R l ibr ary for ge no me -wide as s o cia tion an aly s i s . Bioi n for ma t i cs . 200 7;2 3:1294–6 . 53. Zhou X, Stephen s M . G enome -w ide e ffic i ent mix ed-mo del analy s i s f or a s s oc ia t ion s tudie s. Nat Gene t. Natu re Publi s hing Group ; 2012;4 4 :821–4. 54. Stegle O , P a rts L , Piip a r i M , W i nn J, D urbin R. U s i ng prob abi li s tic e s timatio n of e xpre s si on re sidual s ( PEER) to o bta in inc r e ase d pow e r a nd int erp ret ability o f gene expr es s i o n anal yse s. Na t Pro toc. 2012;7:50 0 –7. 55. M o moz aw a Y, Dmi t rie va J , T héâ tre E, Def fon tain e V, R ahmou ni S, C harlo te aux B, et al . IB D r i s k loc i are e n r i ched i n multige nic r eg ulat ory modul es e ncompa ssing pu t a t i ve c au sat i ve g enes . Na t Commun. 2018;9: 1 –18. 56. Võ s a U , C lari ngboul d A, W e s t ra H, Bo nde r M J , Zeng B, Kirsten H, e t al . Unravel ing the p olyge nic arch itec tur e of c omplex trai ts u sing b loo d eQTL met a- a na lysi s. bioRxiv . 2018 ;1–5 7. 57. A rdli e KG , D eLuc a DS, Seg rè A V. , Sull i van T J , Young TR, Gel fand ET, et a l . The G e no t y pe -Tis s u e Ex pr e ssion (GTEx ) pilot a na lysi s: Multi t i ss ue ge ne r egula t i on in hu ma ns. Sci ence (8 0- ). 2015 ;348:648 –60. 58. Kim-Hellmuth S , Bechh e im M, Pü tz B, Mohamm adi P, N édé l ec Y, Giang rec o N, et al . G e n etic regula to r y ef fec t s modifi ed by imm une ac tivation cont ribut e to a utoimm une di s e as e a ss ociati on s. Nat Commun. Spring er US; 2 017;8 . 59. Sun BB, Ma r a nville J C, Pet er s J E , S tac ey D, S t a l ey J R, Blac k s h aw J , e t al . G eno mic atlas o f th e huma n plasm a prot eome . N a t ur e. Spring er US; 2 018;558:7 3 –9. 60. G ia mbar tolom ei C , Vukc evic D, Sc had t EE, Franke L, Hing or a ni AD, Wallac e C, e t al. Ba ye si an Te s t for Coloc ali sa tion be t w ee n Pa ir s o f G ene tic As soc iation S t ud ie s Using S ummary S t ati s tic s . P LoS Gene t. 20 14;10 . 61. G ia mbar tolom ei C , Zhenl i Liu J , Zhang W, Hauberg M , S hi H, Booc ock J, et al . A Bayesi an framew o r k for mul tiple trait c ol oc aliza tio n from summary a ss ocia t i o n sta tis tic s. Bi oin f o rma t i c s. 2018 ;34:2538 –45. 62. Kim-Hellmuth S , Aguet F , O li va M, M uño z - Aguirr e M, Wuc he r V, Ka s ela S , e t a l . Cell ty pe s pecifi c gen etic r egu lation of g en e ex pre ssion a cr oss huma n t issu e s. bioRx iv. C old Spring Ha r b or Labo rat ory; 2019 ; 63. Roadmap Epigen omic s C on sor t iu m, Kundaj e A, Me ule man W , Ern st J, Bilenk y M, Ye n A, e t al . Integr ativ e an a lysi s o f 111 re fer enc e hu man epige nome s. N a t ure . 2015;51 8:317 –29 . 64. Li MJ, Pa n Z, L iu Z , Wu J, W ang P , Zhu Y , et al. P r e dic ting r e gula tory v arian ts wi th c omposi te sta t i s t ic . Bioin forma tic s . 2 016;32:27 29 –3 6. 65. Kircher M, Wi tt en DM , Jain P , O’ r oa k BJ, C oope r GM, S h endur e J . A g ene ral f ra mew or k for es timating the re lative pa thoge ni city of human ge netic varia nt s. Nat G e n et. N a tu re Publi shing Gr o up ; 2014;4 6:310–5 . 66. Q u an g D, Chen Y, Xie X. D ANN: A de e p learning appr oac h f or anno ta ting t he p athoge nici ty of gen etic v arian t s . Bioi nfo rmatic s. 2 015 ;31:76 1–3. 67. Fu Y, Liu Z , Lou S, Be dfor d J , Mu XJ a s min., Yi p KY, e t al. F unS eq2: a framew o r k for pr i o r i t i z i ng nonc oding r eg ula tory v arian t s in ca ncer. Genome Biol. 2 014;15 :48 0. 68. Shihab H A, R oger s M F , Gough J, M ort M, Coop e r D N, D ay IN M, e t al . An i nte gra tive a ppro ach to predic ting t he fu nc tiona l e f f e c t s of no n-c odin g and coding s e quence vari a t io n . Bio informa tic s . . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint 2015 ;31:1536 –43. 69. Li MJ, Wang LY, Xia Z , Sha m PC, W an g J. GWAS3 D: D e tec ting hum an regula t o ry varian t s by integ r a tiv e ana ly si s of ge nome - w ide a sso cia t i on s , chromo s ome int erac ti on s and h ist one modific ati on s. Nucl e ic Ac id s Re s . 2013;4 1: 150–8. 70. Ryan NM, Mo rri s S W, Por t e o us DJ , T aylor MS, Evan s KL . Su RFing t h e genomi c s wav e: an R pac kag e for p riori ti s in g SNP s by func tion ality . Ge nome M ed. 2014;6:7 9 . 71. Ritchie G RS , D unham I, Zegg ini E, Flic ek P . Func ti onal a nn ota tion o f nonc oding s eq u ence varian t s. Nat M etho ds. 2014;11 :294 –6. 72. Roger s M F , Shi h ab HA, M ort M, Coo p er D N , G a unt TR, C amp bell C. FATHM M - XF : Acc ur a te predic tio n of pa thog e nic point mut ation s via extend ed f ea ture s . Bioin forma tic s . 2 018;34:51 1–3. 73. H ua ng D, Yi X, Zha ng S, Zheng Z, Wang P, Xua n C, e t a l. GW AS 4 D : mul tidimen si ona l anal y s i s of co nt e xt - s pec i fic r e gula tory v arian t fo r hu man comple x disea s e s and t r ait s. Nucle ic Ac ids Re s. O x f o r d Univer sity Pr e s s ; 2018;46: W114 –20. 74. G ul ko B, Hubi sz M J , Gronau I, Sie p el A. A me tho d f or calc ulat ing probabili t i e s of f i t ne ss co ns e qu ence s for poi nt muta tio n s a cr o s s the hum an gen ome . Na t G ene t. N ature Publi shing G r oup; 2015 ;47:276– 83. 75. Chèn eby J, Gh eorgh e M, Ar tu fel M, Mathe l ier A , Ball e ster B. Re M a p 2018: A n updated a tla s o f regula to r y region s f rom an i nte gra tive a n alysi s of DNA-bi n ding ChI P -seq e xpe rime nts . Nucl e ic Ac ids Res . 2018 ;46: D267–75 . 76. Xiong HY, A l ipanah i B, L e e L J , Bre tsch ne ider H, Merico D , Yuen RKC , et al . T he human s pl icing co de revea l s n ew in s ig hts in to th e gene ti c determin ant s o f di sea se . S cienc e (80- ). A me r i can Assoc ia tion for the Adv anc e ment o f Sc ie nc e; 2014; 18. 77. W a ng J, Vic en te -Ga r c ía C , Serugg ia D , Moltó E, Fer nande z - Miñán A , N e to A , e t al. MIR retro t r an spo son seq uenc e s p rov ide in sul ator s t o the hum an genome . Proc Natl A c ad Sc i. 2015 ;112:E44 28–37. 78. Bur r en O S, Rubi o Ga rcía A, J avi erre B M , Rainbow D B, Cai rn s J, Coop er N J, e t al . Chromo some co nt a ct s in ac tiva ted T cel l s iden tify a u t o i mmune dise a se c andid a te ge ne s. Geno me Biol. G enom e Biolo gy; 20 17;18:1– 19 . 79. Freir e- Pri t c he tt P , S choen fe ld er S, Vá rna i C, Winge tt S W , Cairn s J, Colli er A J, e t a l. Glob al reorgani s ation o f ci s- r e gula tory u nit s up on linea ge c ommitmen t o f huma n embry oni c ste m ce ll s . Eli f e . 2017; 6 :1–26. 80. G a spe rini M, H i ll A J, Trapn e ll C, Ahitu v N , Mart i n B , Kim S , et al . Genom e-wi d e framewo rk for mapp ing ge ne regul a tion via ce llula r ge n etic sc ree n s. C e ll. 201 9;176:1 – 14. 81. J a vier re BM, S ewitz S, Ca irn s J , Wing e tt SW, Vár nai C , Thiec ke MJ, e t al . L ineag e-Sp ec ific G enom e Architec ture Link s E nhanc e rs an d N on - c o ding D i sea s e Va r i a nt s t o T arget Gen e Pr o moter s. Cell . 2016 ;167:136 9-1384.e1 9. 82. M i fsud B, Tava r e s-Cad e te F, Yo ung A N, Sug ar R, Sc ho enf elde r S , Fe rreir a L , et a l. Mapping long - range promo ter con tac ts in huma n c ell s with h igh-re sol utio n ca pture H i -C. N a t Gene t. 2015;47 :598– 606. 83. M u mbach M R, Sa tpa thy AT, Boyl e EA , Dai C , Gowen B G , C ho SW, et a l . Enha nc er c onnect ome in primary human c ell s iden tifi e s targ e t g en es o f di sea s e- a s s oc iat ed D NA el eme nts . Nat Gene t. Natu re Publi s hin g Grou p; 2017;4 9:1602 –1 2. . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint 84. Rinaldi L, Da tt a D, S e rr a t J, Morey L, S olana s G, Av gus ti no va A, e t al. Dnmt3a a nd Dnmt3b Assoc ia te w it h Enh anc er s t o Re gula te H u ma n Epide r ma l St em Ce ll Hom eo sta s i s. C e ll Stem C ell. 2016 ;19:491– 501. 85. Rubin AJ , Baraja s B C , Furl an -ma garil M, L opez -pajar es V, Mumbac h M R, How a r d I, e t al. Line age- speci fi c dy namic and pre -e s tabli sh ed e nh ance r – pr omote r cont ac t s c oope rat e in te r mi nal diff er entia tion . N a t G en e t. 2017;49 :1522 –8. 86. W a ng X, He L, Gogg in SM, S aa da t A , W a ng L, Sinnot t-a rmstro ng N , e t al. High - re soluti on g enome - wi de func t io n al di s sectio n of tran scrip tio na l regula t ory r e gion s an d n ucleot ide s i n human. N a t Commun. Spring e r US; 2018; 87. Chen G, Wa ng Z, Wang D , Q iu C , Liu M, Chen X , e t al. Lnc RNADi se a s e : A da ta ba se fo r lo ng-n on- co ding RNA -a ssoc iat ed di sea s e s. Nucl ei c Ac ids R e s . 2013;41: 983 –6. 88. Xie C, Yuan J , L i H, L i M, Zha o G, Bu D, et al . NON CODEv 4: Ex ploring the w orld of long non -co ding RNA gene s . Nucl eic A c id s R e s. 2014;4 2:9 8–103 . 89. D e Ri e D, A buge ssai sa I, A l am T, A rn e r E, Ar ne r P , A s ho or H, e t a l. An integ ra te d expres si on atl as of miRNA s a nd thei r promo ter s in huma n and mo use. N a t Bi o t e chnol . 2017;35 :87 2–8. 90. Bhat t a cha r y a A, Zi ebar th J D , C ui Y. P o lymiRT S Databa s e 3. 0: Li nking polymorp hism s in mic roRNA s and t he i r t a r g et s i te s w ith human di s e a ses a n d biolog ical pa thwa ys. N uc leic Acid s Res . 20 14;42:86– 91. 91. W a ng J, Zh ang P, Lu Y , Li Y, Zhe ng Y, K an Y, et al. pi RBa se : a c ompreh en sive dat aba se o f pi RNA seq uenc e s. N uc leic Acid s Re s . Oxfo rd Uni versi ty Pre s s; 2018;47:1 75–80 . 92. D i xon JR, S elva r a j S, Y ue F, Kim A, L i Y, S hen Y, et al . Topologi cal domain s in m a mmalia n geno me s iden t i fi ed by analy s i s of c h r om atin in te ra ctio n s . Na tur e. N atu re Pu bli s hin g Gr ou p; 2012;48 5:376–8 0. 93. EN C ODE P rojec t Con sortium T, Birne y E, Stam atoyannop oulo s JA, D u tta A, G u ig ó R, Ginge r a s T R, et al . Iden tific a tion an d analy s i s of func ti onal e leme nt s in 1% of th e huma n geno me by the ENC O DE pilo t p r oje c t. N atur e. N atu r e P ubli shing G roup ; 2 007;447: 799–816 . 94. H a r m ston N, Ing -S immon s E, Ta n G, P erry M, Me rken s c hlage r M, Lenha r d B . T opolog ica lly as s o cia ting domain s a re ancie n t f eatu re s tha t coinc ide w it h Met az o an cl u ster s o f extreme n onc oding co ns e rva tion . Na t Commun . S pringe r US; 20 17;8. 95. Rao SS P , Huntl ey M H, Dur and N C, S ta menova EK, Boc hkov ID, Robin so n JT, e t al. A 3 D map o f th e huma n genome at kil ob a se re sol ut i o n re vea ls pr i n ciple s of c h romatin lo op ing. Ce l l. Elsev ie r Inc. ; 2014 ;159:166 5–80. 96. Zieba r th JD , Bhat tac ha rya A, C ui Y. C T CFBS DB 2. 0: a d a t a b a se f or CT CF- bin ding sit es and ge nome organi z a tion. N uc leic Acid s Re s . 20 12;41: D188-94 . 97. Lay er RM, P ede rs en BS, D i s er a T , Mar th G T , G ertz J, Quinl an A R. G I GGLE : a se a rch e ngine fo r large - s c al e in teg r a te d geno m e ana ly sis. Nat M etho ds . 201 8;15:123 – 6. 98. Lande r JM, S upp D M, He H, M arti n LJ , Chen X, W ei r a uch MT, et al . Ana ly si s of chromatin ac ces sibili ty i n human epide rmis id enti fi e s pu ta tiv e barri er dy sfunc t i on -sen sing e nhanc er s . P LoS One. 201 7;12 :1–22 . 99. A l a s o o K, Rod rigue s J, M uk h opadhy a y S , Knig ht s AJ , Mann AL, Kund u K, e t al . Shared gen etic ef fec ts on c hr omatin a n d gene expr e ss i o n indic ate a role f or enh anc er p riming i n immune r e s pon se . Nat . Gen et . 201 8. p. 42 4–31 . 100. Fairf ax BP, Mak ino S , Radhak r i s h na n J, Plan t K, Les lie S, Di lthey A , et al . Ge n etic s of ge n e . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint ex pr e s si on in pr i mary immu ne ce ll s ide nt ifie s ce ll typ e- sp ec ific ma s t er regul ato rs and r ol e s o f H L A all ele s. Nat G e n et . Na tur e Publi shing Gro up; 2 012;44:50 2–10 . 101. Fairf ax BP, Humburg P , Makino S, N aranbha i V, W ong D, L au E , e t al. Inn at e immu ne ac t i vity co ndition s th e e ff ect o f r eg ula tory v arian ts up on monoc yte ge n e ex pr e ssion . S cie nc e (80- ). 2014 ;343. 102. Ishi gaki K, Koc hi Y, Suzuki A, T s uc hid a Y , Tsuchiy a H, S umitomo S, et al . Po lyg eni c burden s on ce ll- s p eci fic pat hwa ys und erlie t he ri s k o f rheumato id a r t hriti s . Na t G e net . Na ture Publi shing G r oup; 2017 ;49:1120 –5. 103. Kasel a S , Kis and K, Tse r e l L, Kalev i s t e E , Remm A, Fi scher K, e t al. Pa t h ogenic implic ation s fo r autoimm une mec h ani sms d erived by c omparative e Q T L an a lysi s o f C D4+ver su s C D 8+ T ce ll s. PLoS Gene t. 20 17;13 . 104. Kim S, Beck er J, B ec hheim M, K ai s e r V, N ou r sadeg hi M, F ric ker N, et al . Cha ra cterizing t h e gen etic ba s i s o f inn at e immune r e s p on se in TLR4-a ctiva t e d h uman m onocy t e s . Na t Commun . 2014 ;5:1–7. 105. Lapp alaine n T, S amm eth M, F r i edlä nde r MR, ’T Ho en P AC, Mon long J, Ri va s MA, e t al. Transcr iptom e and ge nome se quenc ing unc ove r s func t i on al va r i a tion in human s . N a t u re. 2013 ;501:506 –11. 106. Lloy d-Jon e s LR, Hollo way A, McRa e A, Yang J, Smal l K, Z hao J, e t a l. Th e G e n e tic Archi te ctur e of Gene E xpre ssio n in Pe ripher al Blood . Am J Hum Gene t . El sevie rComp a ny.; 20 17;1 00:22 8–37. 107. N a r a nbhai V, Fair fax BP, Mak ino S, H umbu r g P, Wong D, Ng E , et al . Ge nomi c modulator s o f gen e ex pr e s sion in human n eu trophil s . N at Commu n . N a t u r e Publi shing G roup; 20 15;6:1–1 3. 108. N e de lec Y, S anz J, Baha rian G, Tung J, Y otova V, B arrei ro L B, et a l . G ene tic An ces tr y and Na tur al Se lectio n D riv e Popula t i on Dif fe r e nce s in I mmune Re spon se s t o Pa thoge ns . C el l. 2 016 ;167:657– 69. 109. Pala M , Zappa la Z , Mar ongiu M, Li X, D a vis JR, Cu sano R, et al . Po pul ation - an d indiv idual- speci fi c regul a tor y varia t i on in S a rdinia . Nat Gene t. Natu re Publ i s hi ng Gr oup, a d i vision o f; 2017 ;49:700– 7. 110. Raj T, Rothamel K, Mo sta fa vi S, Ye C , Lee M MN M, Repl o gle JM, et a l . Pol a r i z a tion of t h e e f fect s of au toi mmune and n eurod ege ne r a tive r isk a llel e s in l eukoc yte s . Sc ience (80 - ) . 2 01 4;344:519 –23. 111. Q u a ch H, Rotiv al M , Po thli che t J, Ke l s o J, Albe rt ML, Quin tana -murc i L. Gen e ti c Adap t a t io n and Neand ert al A dm ixtur e Sha ped the Im mu ne Sy s t em o f H uma n Popul a t i on s. Ce ll. 2 0 16;167:64 3–56. 112. Sch miedel BJ, Singh D , M adrigal A, V aldov ino- G o nzal ez A G, Whi te B M, Z apar diel -Gonz alo J , et al. Impac t o f Ge ne tic Po lymorph ism s on Human Immu ne C ell G e n e Ex pr e s sion. C el l. 2018 ;1701–15 . 113. W a l s h AM , W h it a k er J W, Huang C C , Cherka s Y, Lambe r th S L, Br odme r k el C, e t a l. In teg rativ e gen omic d econv olution o f rh euma to id ar thrit i s GWAS loc i in to g ene an d cel l type a ssoci a tion s. Genome Bi ol. Genom e Biology ; 2016;1 7: 79. 114. Xia K, Shaba lin AA, Z ou F, Wright F A , Xia K , Sull ivan PF, et a l . see QTL: a s e a r c hable da taba se for huma n e Q T Ls. Bioin forma tic s. 2011;28: 4 51–2. 115. Yao C, J oehan e s R, Jo hn son A D, Hua n T , Liu C, Free dman JE, e t al. Dyna mic R o le of tran s Reg ulatio n of Gene Expre s s i o n in Rel a tio n to Compl e x Trai ts. Am J Hum G ene t. El s e vie rComp an y.; 2017 ;100:571 –80. 116. Ye CJ, F eng T , Kwon H , Raj T, W i ls on M , As in o vs k i N , et al . Int er s ec tion o f pop ul ation va r i a tion and autoimmuni ty genetic s in human T ce ll ac tivatio n. Sci ence (80- ). 2014 ;345:12 54 665. . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint 117. Zhernak ova D V., D e elen P , V ermaa t M, Van I t e r s on M, Van Ga len M, Arindr a rto W, et al . Identi fic a tion o f c ont ext-d epe nd ent ex pres sio n quan t i ta tiv e tra it loc i in w hole bl o od. Na t Ge net . 2017 ;49:139– 45. 118. Fag ny M, Pa ul s o n JN , K uijjer ML, So naw ane AR, Chen C -Y, L ope s- Ramo s CM , et al . Exp lor i ng regula ti on in ti s s u e s w ith e QTL ne twork s. Proc Na tl Ac ad S ci. 2 017;114 :E 7841–50 . 119. Andiappa n AK, Melc hio tti R , Poh TY, Nah M, P u an KJ, Vig ano E, et al. G enom e -wide analy s i s o f the g ene tic r eg ula tion o f gene e xpre ssio n in human neutro phil s. Na t Commun. N atur e Publ i s hi ng Gr o up ; 2015;6 :7971. 120. Bat tle A, M o st a favi S , Zhu X, Pot a sh JB, W eis s ma n MM, Mc Cormic k C, e t al. C ha r a ct eriz i ng th e gen etic ba s i s o f tr an scriptom e div er s ity th r oug h RNA- seq uen cing of 922 i ndividu a ls . G e nom e Re s. 2014 ;24:14–2 4. 121. Bonde r MJ, Lu ijk R, Zhernako va D V., Moe d M, De elen P , V ermaa t M, e t al. Di se as e va r ia n ts alter tran scri ption fac t or leve l s a nd me th yla tion of thei r bin ding si te s. N a t G e n et. 2017;49 :131–8. 122. Thompson PJ, Mac f arlan T S, L or i ncz MC. Lo ng Termina l R epe a ts: F rom Par a s i t ic Ele ment s to Build ing Bloc ks o f th e Tran sc r ip tio nal R e g ulatory Rep e r t oire . Mol Cell. El s e vie r In c .; 2016 ;62:766– 76 . 123. Page a ud Y, Pla s s C , A ssenov Y. Enric hment ana ly si s w ith EpiAnno t a t or . Bi oin f or m atic s. 2018 ;34:1781 –3. 124. G u t i e r r ez -Ar c el us M, La ppala in en T, M on t g omery S B, B uil A, Ong e n H, Yurov sky A , et al. P assive and activ e D N A me t hy lation a nd th e inte rpla y wi t h genetic va r i a t i on in g en e regul ation . Eli fe . 2013 ;2013:1– 18. 125. D i ng J, Gudjon sso n JE, L iang L , Stua rt PE , Li Y, Chen W, et al . G ene expr e ss i o n in skin a nd ly mphoblas toid c ell s: R efi ned s ta ti st i cal method r evea l s exten sive ove rlap in ci s - e QTL s ig nal s. Am J Hum Gene t. 20 10;87:779 –89. 126. Min JL, H em ani G, Hanno n E, Dekk e r s KF , C a stillo -F erna nd ez J, Luijk R, et a l. Genomic and pheno mic i ns ig h ts fr om an atla s o f gen e tic eff ec ts on DNA me thyl ati on. medRx iv. Cold Spring Ha rbor La borat or y Pr es s; 20 20; 127. Emils s o n V , Ilk ov M, L amb JR, F inkel N , G udm und sson EF , Pit t s R, e t al. Co- r eg ulatory ne tw or k s of huma n s e rum pr o tei n s l ink g enetic s t o dise as e. S cie nc e (80 - ) . 2018 ;36 1:769–73 . 128. Suhre K, Arnol d M, Bh agw at AM, Co tton R J, Eng elke R, R af fle r J, e t a l. Conn ec ting g enetic r i s k t o dise a se en d p oint s t h roug h th e human bl ood pl a s ma pro t e ome . Na t Commun . 2017;8. 129. Pelik an R C, K elly J A , Fu Y, Lareau CA, Tes s ne er KL, Wil ey GB , et a l . Enha nce r hi st one - Q T Ls a re enric hed on a utoimmun e ri sk ha plotype s and influenc e ge ne ex pre ss ion wi t hi n ch romatin n etwork s. Nat Commun. Spring er US; 2 018;9 . 130. Mauran o MT, H a ugen E, Sand s trom R, Vie r st r a J , Sha fe r A, Ka ul R, e t a l . L arge - s c ale iden t i fi cati on o f s e que nce vari ant s in flue nc ing h uman tran scrip tion f ac to r occ upa ncy in v ivo. Nat Gene t. 20 15;47 :1393 –401. 131. Cole C, Kr o bo t h K, Sc hurch N J, Sa ndi land s A, S her stnev A, O’Rega n G M, e t a l . Fi lagg r in -st r a ti fi ed tran s c rip tomic analy s i s of p ed iat r i c skin i denti fie s mec hani s t i c pa t hw ay s in p at ien ts with atopic dermati ti s. J Allergy Clin I mmunol. E l s ev i er Inc . ; 2014;1 34:82–9 1 . 132. Ewa ld DA, Mal aji an D, Kru ege r JG, Workma n CT, W ang T, Ti an S, et al . Me ta -a naly si s derive d atopic de rmati tis ( MA D A D ) t ran scrip to me de f i n e s a robu st AD signa t ure highlig hting the inv olveme nt of athe ro s c ler o si s a nd lipi d metaboli sm pa t hw ays . BM C Med Gen om ics. BM C M edic al Genomic s ; 2015;8 :1– 15. . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint 133. G h o s h D , Ding L, Siva pr a s ad U, G e h E, Bia gini J. M ul t i ple Tra nsc riptom e D a t a Analysi s R eve al s Biolo gica lly Re leva n t A to pi c Derm ati tis Si gn atur e G e n e s and Pa thw ay s. PL oS One . 2015 ;10:e014 4316– e0 144316. 134. Sää f A M, Tengv all -L inder M, Chan g HY, Adler AS , W a hlgr en CF , Sc h eyni us A, e t al. Globa l ex pr e s ion pr o f i ling i n at opic eczema rev e als r eci pr o cal e xpre s s ion o f inflamma t o r y and li pid ge ne s . PL oS One . 20 08;3. 135. W i nge MC G, Hop pe T , Be rne B, V a hl quis t A, No r de n s k jöld M, B radle y M, et a l . Fi laggrin gen ot y pe de te rmine s func tional a nd mol ec ular a l t e r a t i on s i n s k in of p a t i e nts w ith atopic derma titi s and ichthyo s i s vul garis . PLoS On e. 2011 ;6 :1–8. 136. Q u rai shi BM , Zhang H , Everson TM , Ra y M, L ock ett G A, Hollow ay JW, e t al. Id e nt i fy ing Cp G sit e s as s o cia ted with ec z em a via rand om for e s t s c re ening o f epigen ome -s c ale D NA me t hyla t i on. Clin Epig enetic s . Cl inica l Epig ene tic s; 20 15;7: 1–11 . 137. Rodríguez E, Ba urech t H, Wah n AF , Kret s c hmer A , Hotze M, Zeili nge r S, et al . An i nt e gra ted epig ene tic and tr an script omic analy s i s r e ve als di sti nct ti ssu e- sp eci fic pa t t ern s of DN A m e t h ylati on as s o cia ted with atopic d ermati ti s. J Inve st De rmatol . 2 014;134:1 873–83 . 138. Elias MS , Long H A, N ew man CF , Wil son PA, W e st A, McGill P J, e t al. Pro teom ic ana lysi s of fila ggrin de fic ienc y id en tifi e s molec ular si gnature s cha r a ct eri s tic o f at opic ec z ema . J Allergy Clin Immunol . Els evi er Inc .; 2 017;140: 1299 –3 09 . 139. Molin S , Me rl J, Die trich KA, Regau e r M, Flaig M, Le t u l é V, et a l . The hand ec ze ma prot eome : Imba lance o f epid e r ma l ba rrie r pro tein s. Br J De r ma to l. 2015; 172:994 – 1001. 140. Milano A, Mc Mahon A , W e l t e r D, B o wler E, H a s ting s E, Cunni ngha m F, et al . Th e new N HGR I-EB I Cat al og of publ i s h e d genome -wide a ssoc ia tion st ud i e s (GWAS Ca t a log) . Nucl eic A c ids Re s . 2016 ;45:D896–90 1 . 141. Bogunie wic z M, L eung D Y M. Atopi c de r ma t i ti s: a di s e a s e o f alt ered skin ba rr i er and immu ne dysregul a t i on . Immunol Rev. 2011;242 :2 33–46. 142. D ri skel l RR , L ichtenb e r g er B M, Ho s t e E, Kretzschma r K, Simon s B D, C ha r a l ambous M , et al . Di st in ct fibr obla st lin ea ge s de termin e de r ma l arch ite ctur e in skin d evel opmen t an d r ep air . Na ture . Natu re Pu bli shing G rou p; 2013;504 :277 – 81. 143. H i ll AB. Th e e nvi r on men t and di s ea se: a ssoci ati on or ca u s a t i on? J R S oc Med . 1965;58: 295–300 . 144. Simeo nov DR, Gowe n B G , Bo ontan r art M, R oth TL, G a gnon JD , Mumba ch M R, e t a l . Di scove r y o f st i mul ation -r e spon s iv e immu ne en ha nce rs with CRIS PR ac t i va tion . Na ture . Na tur e Publ i s hi ng Gr o u p; 2017 ;549:111 –5. 145. W a inbe rg M, S inno tt -A rm s tr ong N, Mancu so N, Ba rbei r a AN, K nowle s DA , G olan D, e t al. Oppor t un it i e s a nd c halle ng e s f o r tra nsc ri ptome -w ide a s s o cia tion s tudi e s. N at G en et. S p r in ger US ; 2019 ;51:592– 9. 146. Mif s ud B, Tav ar e s - Cad ete F, Y oung AN, Sug ar R , Sc hoen fel d er S, F e r rei ra L, e t a l. Map pi ng long - range promo ter con tac ts in huma n c ell s with h igh-re sol utio n ca pture H i -C. N a t Gene t. N a t ur e Publi s hin g Grou p; 2015;4 7:598 –60 6. 147. Espa rz a -Go rdillo J , Scha ar s c hmid t H, Lia ng L , Cookson W , Ba ue rfe ind A, L e e-K irsch M A, et a l. A func t i o nal IL -6 r ecep tor (IL 6R) vari a nt i s a ris k f actor for p er si s t ent a topi c de rmati t is. J Alle rgy Clin Immunol . 2013; 132:371 – 7. 148. G a r b e rs C , Monh a sery N , Ap aricio -S i egmund S, L okau J, Bar an P, Now ell MA , et a l. The interl euk in-6 rece p t o r A sp358 Ala s i ngle nu cleotid e polymo r phi s m r s222814 5 con fer s inc rea s e d . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint prote olytic co nver s i on r ate s by A DAM p rotea se s. Bioc him Bi op hys Ac ta - M ol Ba s i s Di s . Els evi er B. V .; 2014 ;1842:14 85–94. 149. Rosa M, C hign on A, Li Z, Boul ange r M- C, Ar s enaul t BJ, B os sé Y, e t a l . A Me nd el ian randomiza tion study o f IL6 sig na ling in ca r d iovascul ar di sea s e s , immune -rel at ed di s or der s and long evity. npj Genomic M ed. S pr i ng er US ; 201 9;4:1–10 . 150. In s h aw JR J, Cu tle r AJ, B urr en OS, S te fana M I, Todd J A. App roac h es and adv an ce s i n th e ge ne tic ca use s o f a u toimmune di s e a s e and their impl ication s . N at Immu nol . Spring e r U S ; 2018;1 9:674–84 . 151. Kalam aha K, Rei s E, N ewton S , Roc h e C, Jul s on J, F ernand e s H, et al. A t opi c d erma t i ti s: a rev iew of ev olvin g targe ted ther api e s. Expert Re v Clin Im munol. Ta ylor & F ranc i s ; 2019;1 5:2 75–88. 152. Lin C-T, Huang W- N, H s i eh C - W, Ch e n Y - M, Ch en D -Y, H sieh T -Y, et al . S a fe ty a nd ef fec tiv ene s s of toc ilizu mab in t r eatin g pa tien t s with r heu matoid arth riti s – A thre e -year s tudy in Taiw an. J Mic r obi ol Immunol I nf ect. 2 019;52 :14 1– 50. 153. W i t tm an n M, P urwar R, H a r tman n C , Gutz m er R, Wer fel T. Human Ke ra t i n ocyte s Re spon d to Inte rleuk in -18: Imp lica tio n for the C our se of C hronic I n flamma tory S kin Di se a se s. J Inve st Derma tol. Elsev ier ; 2005;1 24:1225– 33 . 154. H e dl M, Zh eng S , Ab r a ham C . The IL 18RAP Reg io n Di se as e Poly morph i s m D e c rea se s IL - 18RAP /IL-1 8 R1/ I L -1R1 Ex pre ssion a nd Si g naling thr o ugh In na te Rece p t or - Initia t e d Pathwa y s . J Immunol . 2014; 192:5924 –32 . 155. Soskic B, Cano - Game z E , Smyth DJ, Row an WC, Nak ic N, E sparz a- G o rdil lo J, e t al . Ch roma tin ac tivity a t GWAS l oci i den ti fie s T c ell s tat es d r i ving comple x immun e disea s es. bio Rxi v. 201 9;566810 . 156. Mumbac h MR, Sa tpa t hy AT, Boy le EA, D a i C, G owen BG, Cho S W, e t a l. Enha ncer c onnec tome in primary human c ell s iden tifi e s targ e t g en es o f di sea s e- a s s oc iat ed D NA el eme nts . Nat Gene t. Natu re Publi s hin g Grou p; 2017;4 9:1602 –1 2. 157. Al-mo ssaw i H, Yag er N , Tayl or C , Lau E, Danie lli S , Wit D, et al . C o nte xt- s p ec ific r e gula tion o f monoc yte s urfa ce IL7R e xpr es s i on and so luble r ec e ptor se cre tion by a c ommon au toim mune ri sk all ele . bi oRxiv . 2018; 158. G a la rz a - Muñ oz G, Brigg s F BS, E vs y u kov a I, S chott -Lern er G, Kenn edy EM, N y anhet e T, et al . Human E pis t atic In te rac t io n Con tr o l s I L7 R S plici ng and Increa s e s Multipl e S cle r o si s Ri sk. C el l. 2017 ;169:72-84 .e13. 159. G i an felic i V, Me s s i na M , Pa oloni F , P eragi ne N, La ur e tti A , F edullo AL, et a l . IL 7R overex pre ssion in a dult ac ute lymph obla st i c le uk emia is assoc ia ted to J AK/STAT pa t h way mutatio ns and id e nti fie s patie n ts wh o coul d bene fi t fr om targe te d t he r a pie s. Leu k Ly mphoma. Tayl or & F ran cis; 2018 ;60:829– 32. 160. Puel A , Ziegl er SF, Buckl ey RH, L eo n ard WJ . D ef ectiv e IL 7R ex pr e s sio n in T -B+ N K+ se vere co mbined i mmunode fic ienc y. Na t G ene t. 1998 ;20:394– 7. 161. Kicha ev G, Bha t i a G , Lo h P, G azal S , Burch K, Freund MK, e t al . L everag ing Pol ygeni c F unctiona l Enric hment to Improve GWAS P ower . A m J Hum Gene t. El s e vier Compa ny .; 201 9; 1–11. 162. G a n X, Wang J , Wang C , Sommer E, Koz a sa T, Srini vas ula S , e t al. PRR5L degr adation p r om ote s mTORC2-m edia t e d PK C-δ ph o sphoryl a tio n and cell migration do wn s t re am of Gα 1 2. Nat Cell Bi ol. Natu re Pu bli shing G rou p; 2012;14:6 86 –9 6. 163. Xing Y , Iga r a s hi H, Wan g X, S akag uc hi N . Prot ei n pho spha ta se s u buni t G5 PR i s ne ede d for inhi bition o f B c ell r ec ept or – induc e d a p opto s i s. J Ex p Med. 2 005;202:7 07–19 . . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint 164. Xing Y , Wang X , Igarashi H , Ka wamoto H, S aka guchi N. Pro t e in ph o sphat a se s ubunit G5 PR th at regula te s the J NK -medi a ted ap opto si s s ig nal is e s s ential for the surviv al of CD4 a n d CD8 doubl e - posi tive t hy mocy t e s . Mol Immunol. 2008 ;45:2 028–37. 165. Elling haus D, Jo stin s L , Spai n SL , Co r t e s A, B ethun e J, Ha n B, e t al. An a lysi s o f five c hronic infla mmat ory dis ea se s id enti fie s 27 new a ssocia t i on s a nd hig hligh ts di s ea se - s pec i fic pat tern s a t sha red loc i. Nat Gen et . Na ture Publi shin g G roup; 2016;48 : 510. 166. Astle W J, E lding H, Ji a ng T, A l len D , Ruk lisa D, Mann AL , et al . The a llelic la n d s c ap e of hu man blood cell tra it varia tion and li nk s to common comple x disea s e. C e ll. El s ev ie r; 201 6;1 67:1415– 29. 167. Chiang C -WC -W, L iu W -K, Chi a ng C - WC -W, Chou C -K. P ho s pho ry lation -d epe nd ent a s s oc ia tion o f the G4 -1 /G 5 P R regula tory subuni t w ith I KKβ n egativ ely mo dulat e s N F -κB ac tivati on t hroug h recruitm ent o f pr o tei n ph os phat a s e 5. Bi oc hem J. 2010;433 :187 –96. 168. Mahmudul F, Ronny H, Iga ra shi H, S akag uchi N . BC R-cr o s s link ing ind uce s a tr anscri ption o f prote in pho spha ta se c ompon ent G5PR t ha t i s requi r e d for mat ur e B -cel l s urv ival. 2006;34 0:338–46 . 169. Kitabat ak e M, Tod a T, Ku waha ra K, I ga r a shi H , Oh tsu ji M, Tsur ui H, et al . Tran sgenic Overexp r e s sio n of G5PR T ha t I s N o rmall y Augme nt e d in Ce n t roc yt es Impair s th e Enric hment o f High- Af finity An tig en -Spec i fic B Cell s , Inc r e a s e s P e r it one al B-1a Cell s, and I nduce s Auto immuni t y in Age d Fe male Mic e . J Immu nol . 201 2;189:119 3 –201. 170. Altr ic h ter S , Krie hub er E, Mo se r J, V alent a R, Kopp T , Stingl G. Serum Ig E Au t oantibo die s T arge t Keratin oc yte s in Pa t i en t s w ith Ato pic De r matiti s. J Inv e s t Derm atol . E lse vier Ma s son SAS; 2008 ;128:223 2–9. 171. Li F , Li u X, Zh ou W, Yang X, S hen Y. A u t o -in hibi tory mec hani sm o f the h uma n mitoc hondrial RNa se P prot ein c omplex . Sc i Rep . N a tur e P ubli shing G roup; 201 5;5 :9878. 172. Shen J , Q i a o Y, Ra n Z, Wang T. Up -r e gulatio n and Pre -ac tivation o f TRAF3 and TRAF5 in Infla mma tor y Bowe l Di s e a se . In t J M ed S c i. 20 13;10:156 –63. 173. Xu Y, C heng G, Ba l timore D . Target e d dis rup tion o f TRAF3 l e ad s to po stna ta l l ethal i t y and def ec tive T -d epend ent i mmune re spon se s. I mmunity. 19 96;5 :407– 15. 174. Bish op GA, Stunz LL , Ho s t age r BS . TRAF 3 a s a mul t i fa cet ed r eg ulato r of B ly mphocy t e s u rvi val and activ atio n. Fron t Imm unol . 201 8;9:1 –10. 175. Tseng P , M a tsu zawa A, Zhan g W, Mi no T, Vignal i D AA , Karin M. Di f fer ent mo des o f ubiqu it i na tion o f th e a dap tor TRAF3 s el e ctive ly ac tivate the expre s s i o n of type I i nter fe r o n s a nd proinfl am mato r y cy t o kines . Na t Immuno l. 2010;11 :70–5 . 176. Ma C, L in W, L iu Z, Tang W , G a u t a m R, Li H, et al. N DR1 pr ot ein kina se promo te s IL ‐ 17 ‐ and TNF ‐ α ‐ media ted in flamma t i o n by c ompetitive ly bi nding T RA F 3. EMB O Rep . 2 017; 18:58 6–602. 177. Fine RS, P e rs TH, Amar iuta T, Ray ch audh ur i S. B enc hmarke r /i1: An Unbi a sed , As s oc ia tion -D ata - Driven S t r a teg y to Ev alua te G e n e Pri orit i z a t i on Algo rithm s. Am J Hum Gen et . Els e vi erCompany. ; 2019 ;1–15. 178. Kulesh ov M V, Jo ne s M R, R ouilla r d A D, F e rnand ez NF, Duan Q , Wang Z, e t a l . Enric hr: a co mpr e hen sive g en e s et e nric hmen t ana l ys i s w eb s erver 2016 upda te . Nuc leic Aci ds Re s . 2016 ;44:W90 -7. 179. Esaki H, Br unne r P M , R ener t - Y uval Y, Czarn owic ki T , Huynh T , Tran G, et a l . Early -on set pe di at ric atopic de rmati tis i s TH2 but al so T H 17 polarized in ski n. J Allergy Clin Im munol. 20 1 6;138:16 39–51. 180. Suárez -Fa riña s M, Dhing r a N , G i ttl er J, S hemer A, C ardin al e I, d e Guz ma n S tro ng C, et al . . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint Intrin sic a t opi c de r m a t i tis show s similar TH2 a nd high er TH17 immune activ ation c ompared w ith ex t rin sic atopic d ermati ti s . J Allergy Clin I mmunol . 2013; 132:361– 70. 181. Cz a rnow ick i T, G onzale z J, Bon i faci o KM , Sheme r A , Xiang yu P, K unjravia N , e t al. Div e rse ac tivation an d dif fer enti atio n of multi ple B- c ell s ub se t s in pati ent s with a topic d e rma t i t i s but n o t i n patie n ts wi t h p s oria si s . J All ergy Clin I mm unol. 20 16;137:1 18 -129. e5. 182. Szklarc z yk D, Ga ble AL , Ly on D, J unge A, Wyder S , H ue rta -c epa s J, et al . STRI N G v11: prote in- prote in a ss ocia t i o n netwo rks wi t h incr ea sed c ove rag e, suppo rting func t i on al di sc overy i n genome - wi de ex periment al da ta s e ts . Nucle ic Ac id s Re s. O x fo r d U niv er s ity P re ss; 2019 ;47 :6 07–1 3. 183. Jeng MY, M u mbach MR, G r anj a JM, Sa t pa t hy AT, Cha ng HY, Ch ang AL S. Enh a ncer Co nn ectom e Nominat e s Targe t G en e s of I nheri ted Ri s k Varia n ts fr om Infl am mato r y Sk in Diso rder s. J Inv e s t Derma tol . Th e Auth ors; 20 19;139 :60 5–1 4. 184. Mucha S, Bau rec ht H, N o vak N , Rod r í guez E, Be j S, May r G, et al . P rot ein -c od ing variant s co nt ribu te to t he ri s k o f at opic derma titi s and skin -s pec ific g en e ex pre ssion . J A l l e rgy Cl in Immunol . Americ an Aca d emy of A l l ergy, A sthma & Immunol ogy; 2 019; 185. Zhu BM, I shida Y, Robi n s o n G W , Pa c her -Zav isin M, Yo shimur a A, M u rphy PM, et al . S OCS3 neg atively regula t e s t he g p130-ST AT3 pa thw ay in mouse s k in w ound he a ling. J I n vest D e r m a t ol . Elsev ier Ma sson S AS ; 2008 ;128:1821 –9 . 186. Car ow B, Rotten b er g ME. S OCS 3, a ma jor r e gulat or o f inf ec tion an d in flamm a t i on . Front Immunol . 2014; 5:1 –13. 187. Moi s a n J, Gre nningloh R, B et telli E, Oukka M, Ho I -C . Et s- 1 i s a n eg ative r eg ulator o f Th17 diff er entia tion . J Exp M e d. Uni ted S tat e s; 2007 ;204:2825 –35. 188. Frobø se H , Gr o th Røn n S, Hedin g PE , M e ndoza H , Coh en P , Mandru p- Poul se n T, et al. Sup pr e s sor o f Cy tokin e Sig naling - 3 Inhibi ts Int erleuk in -1 Sig naling by T argeting t h e TRAF-6 /TAK1 Comple x. Mol Endoc rinol . 2006; 20 :1587 –9 6. 189. Pari si en J, L eno ir J J, Mandh a na R, R odrig uez KR, Q i an K, B run s AM, e t al . R NA s e ns or LG P2 inhi bits TRAF ubiqui tin liga se to n e gative l y regu late inn a t e immune s ig naling . EMB O Rep. 2018 ;19:e451 76– e451 76. 190. Li L , Zh ou D, Zheng Y, Xi e W. Ex pr e ss ion an d func tion s o f the S TAT3 -SCL IP pa t hw ay i n chronic mye loid le ukemia cel l s . Exp T her Med . 2 01 6;12:3381 –6. 191. N g DCH, Bao HL, Ch eh PL , Hu ang G, Z hang T , Poli V, e t al. S tat3 r egula te s mi crotubul e s by antag onizing the de p olyme r iza tion a ctivi t y of st a t hmi n. J Cell Bi ol. 20 06;172:2 45 – 57. 192. N g DCH, Lim CP , Lin BH , Zhang T, Ca o X. SC G 10 -l ike pro tein (SCL IP ) is a STAT 3 -int er a c ting prote in in volve d in ma intaining e pith elia l morpholog y in MCF -7 brea s t canc er ce ll s. Bioch em J . 2009 ;425:95– 108. 193. Mahaja n A, Tal iu n D, Thurn er M, R o be r t son NR, Tor re s J M , Rayner NW, e t al. Fine-ma ppi ng typ e 2 di abet es l oci to s i ngle -va r i an t re s ol u t i o n us i ng hig h-d en sit y imput ation a nd isl et - s pe ci fic epig enome map s. N a t Gen e t. 2018;50 :15 05–13 . 194. Canno n ME , Mohlke KL. D ec ipheri n g t h e Emergin g Comple xiti e s of M olec ula r Mec hani sm s at GWA S Loc i. Am J Hum Gen e t. Else vierC o mpan y.; 201 8;103:63 7–53 . 195. W o od AR , Hern andez DG, N a ll s M A , Yaghootk ar H , G i bb s JR, H a rrie s LW, e t al . Alle lic hete roge n eity an d mor e det ail ed an a lys e s o f known loci expla in add itio nal ph en ot ypic variatio n and revea l c omplex pat terns of a ssocia t i on . Hum Mol Gen et. 2 011;20 :40 82–92. . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint 196. Bonora G, P la t h K, Denhol tz M. A mec hani st i c li nk betw een ge n e regula t i on and g enome arch itec tur e in ma mmalia n dev elopm ent . Cur r Opin G ene t D ev. Els evier; 2 014;27 : 92 –101. 197. Mor l ey M, M olo ny CM, Web er T M , D e vlin JL , Ew en s KG , Spiel m an RS, e t al. Gene t i c ana ly si s of gen ome-wid e variati on in huma n gene e xpres s i on . Na ture . Na t u re Publi sh ing Gro up; 20 04;430:74 3. 198. Cor r a din O , Sai akhov a A, Akhta r-Za i di B, My ero f f L, Willi s J, C owp er -Sal la ri R, et a l. Combin a tor i al ef fec ts of mul t i ple enh a nc er v arian ts in link ag e d iseq uilibri um di cta te le vel s o f ge ne ex pr e s si on to c on fe r s u sc e pt i bili ty to c o mmon trait s . Gen ome Re s . 20 14;24:1– 13 . 199. N a gy R, W a ng Y, Shen R, He H , Akag i K, L iyana r a chc hi S, et al. Multipl e func ti ona l va r i ant s i n long -r a ng e enh anc er e l ement s con tribu t e to th e ri sk of S N P rs9655 13 in thy r oi d can cer. Proc Na t l Aca d Sc i. 2015 ;112:612 8–33 . 200. Tsoi LC, R odriguez E, Deg enh ard t F, Vol k s N , Sz y mczak S, Sw indel l WR , et al . A topi c D erma titi s I s an IL-13 D omin ant Di s ea se with Gr eat er Molec ular H et er o gen eity C ompare d to P sori a si s . J I nves t Derma tol . 20 19;139:14 80–9. 201. Br u nne r PM , Gut t ma n -Ya s s ky E, L eung D Y MM . The im munology of a topic de r matiti s and i ts rever sibili ty with broad - spec t rum a nd t ar geted t hera pi e s . J Alle rgy C lin Immuno l. 2017; 139:S65 –76. 202. Imam T , Park S , Kapl an M H, O l s on MR. Ef fect or T Helpe r C ell S ub s et s in I n f l ammatory Bow el Di sea se s. Fron t Immunol. 2018 ;9:12 12. 203. Sega l BM. S ta ge -Spec i fic Immune D ysregula t i on in Mul tipl e Sc l ero s i s. J In ter f Cy t ok ine Re s. 2014 ;34:633– 40. 204. W a lker L SK, v on Herra t h M . C D4 T c ell di ffe ren t i a t i on in typ e 1 dia be te s. Clin Ex p Immuno l. 2015 / 07/28. J o hn Wil ey and So n s In c. ; 201 6;183:16 –29. 205. Canno n ME , Dua n Q , Wu Y, Z eynal zade h M, Xu Z , Kang as AJ, e t al. T ra n s -an c es t ry Fine Ma pping and M o lecul a r A ssays Iden tify Regul a to ry Varian ts at th e AN GPT L8 HDL-C GWA S L ocus. G3 ge ne s - gen ome s - Ge ne t. 2017;7 :3217–27 . 206. Viei ra Brag a F A, Ka r G, Berg M , Car p aij O A, Pol an ski K, Simon LM, et a l . A ce ll ular c en su s of huma n lungs ide n t i fi e s n ovel c ell s t at es i n hea lth an d i n as t h ma . Na t Med . 20 19;2 5:115 3–63. . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint Table 1 Genes prioritised by ADGAPP at atopic dermatitis GWAS loci The close s t gene s t o the in d ex v ariant (in ei t h e r directi on) a r e marke d i n bold . Th e two v alue s giv en in paren the se s i n top 3 rank ed gen e col umn s corre s pond t o t h e ADGAP P gene p r i or i ti sati on sco re and p e rc entag e of t he to t a l scor e fo r loc u s top 10 gene s, re spec t i vely . L o c u s G WAS I nd e x v a r i a n t N e a r e st g e n es To p r an k ed g en e 2n d r an k e d g e ne 3 r d r an k ed g en e 1q 21.3 - a rs6181 3 87 5 CRC T1/LC E 3E HRN R ( 46 4, 2 8%) RP T N ( 2 8 5 , 17% ) CRN N ( 2 49, 15% ) 1 q2 1 . 3 - b r s 12 7 309 35 I L6R IL6R ( 74 3 , 6 2% ) U BE 2 Q1 (9 3 , 8 % ) ADA R (61 , 5 % ) 2 p1 3 . 3 r s 11 2 111 458 C D207 /VA X2 C D 207 (272 , 4 5 % ) C LE C4F (62 , 10 %) V AX2 ( 56 , 9 % ) 2 q1 2 . 1 r s 64 1 957 3 / rs 39 172 65* I L1 8R1 / I L 18 RA P I L1 8R1 (138 4 , 39 % ) I L 1 8R AP (134 1, 3 8%) IL1RL 1 (2 2 4, 6%) 2 q3 7 . 1 r s 10 572 58 I N P P5 D INPP5D ( 2 96 , 57% ) ATG 16L 1 (1 0 6, 20% ) RN 7S L 32P (29, 6 % ) 4 q27 rs682 77 5 6 / rs1 31 5 2362* KI A A 110 9 K I A A11 09 (220 , 3 5%) BB S1 2 (1 12 , 18 %) TR P C3 ( 10 0, 16% ) 5 p1 3 . 2 r s 10 2 142 37 I L7 R /C A PS L IL7R ( 96 5 , 6 5% ) S P EF2 (2 0 3, 14 %) UGT3 A 2 (8 9, 6% ) 5q 31.1 - a rs1 218 8 91 7 T H 2L CR R SLC2 2A 5 (461, 3 5 %) I RF 1 ( 30 3 , 2 3 % ) R AD 50 ( 122, 9%) 5q 31.1 - b rs4 705 9 62* KI F 3 A K I F 3A ( 2 49, 23% ) S LC 2 2A5 ( 247 , 23 %) P D L I M4 (1 42 , 13 % ) 6p21. 32 rs4 7 1 3 55 5 STA T 3 HLA- D RA ( 1 405 , 3 0%) HLA- DQ B 1 (6 8 9, 15 %) H LA - D R B1 (566 , 12 % ) 6p21. 33 rs4 129 3 864 M IC B HS P A 1 B ( 1 73, 15% ) HCG 27 ( 1 6 5, 14%) CS N K2 B ( 15 2, 1 3% ) 8q 2 1 .1 3 r s 64 732 27 M I R5 708 / Z BTB1 0 Z BTB1 0 ( 19 2, 4 1 %) TPD 5 2 ( 70, 15 %) P AG 1 (69, 15% ) 10 p 15 . 1 r s 66 023 64 I L 2 RA / I L 15R A I L 2 RA (3 33, 45%) RB M17 (111 , 15 %) P F K FB 3 ( 51 , 7 % ) 10 q 21 . 2 r s 29 445 42 Z NF 365 AD O ( 61 5 , 61% ) ZNF 36 5 ( 10 1, 10 % ) E GR2 ( 90 , 9%) 11p 13 r s 25 92 555 /r s 12 295 5 35 * P RR 5 L P R R 5L ( 5 98 , 79% ) T RAF 6 (6 5, 9%) COM MD 9 ( 34, 5 % ) 11 q 13 . 1 r s 10 7 918 24 O V O L 1 C T S W (3 36, 23 % ) OVOL 1 (2 36 , 16% ) EFE MP2 ( 168, 1 1 %) 11 q 13 . 5 r s 22 124 34 C 11 o r f 30 / L RR C3 2 L R RC 3 2 (5 45, 43 %) E MSY (521 , 41 % ) T H AP 1 2 ( 47 , 4%) 11 q 24 . 3 r s 71 273 07 – / ET S 1 E TS 1 (29 8 , 7 5%) F L I I (3 5, 9 %) A P L P2 ( 18 , 5 % ) 12q 15 r s 22 274 83 I L 2 2 M DM1 (728 , 70 %) I L 22 (9 9, 1 0% ) I F N G (57 , 5% ) 1 4q1 3.2 rs2 0 3 8 25 5 PPP2 R3 C P P P2R3 C (9 96, 31 %) K IAA 03 91 (81 4, 2 5 % ) S RP 54 (43 3, 13 % ) 14 q 32 . 32 r s 71 465 81 TR A F3 T RA F3 (84 8 , 5 5% ) AM N ( 2 81, 18%) CD C4 2BP B ( 18 6, 1 2 % ) 16p 13 . 13 r s 20 417 33 C L E C1 6 A DE X I (37 6, 34% ) C LE C 16 A (3 64 , 33 %) RMI2 ( 1 0 8, 1 0%) 17 q 21 . 2 r s 12 9 519 71 S T AT3 D HX58 (2 54 , 3 2 %) ST AT3 ( 10 1, 13 % ) RA B 5C (1 00 , 13 %) 17 q 25 . 3 r s 11 6579 87 P G S1 PG S 1 ( 20 5, 4 6% ) D NA H17 ( 73 , 1 6%) S OC S 3 (52 , 12 % ) 19 p 13 . 2 r s 29 183 07 ADA M T S1 0 /A CT L 9 AC TL 9 ( 1 1 5 , 41 %) ADAM T S 1 0 ( 5 7, 2 0 % ) M A P 2 K 7 ( 3 4, 1 2% ) 20 q 13 . 33 r s 48 092 19 R TE L 1 /T N F RS F 6B S T M N3 ( 6 08 , 27% ) LI M E 1 (4 73 , 21% ) AR F R P1 (2 57 , 12 % ) * i n de x SN P f o r s ec ond a r y si gna l, w h ere AD GA P P d i d not giv e diff ere n t gen e prioriti sa tion s f or t he t w o si g n a ls, t he s e are pre sente d on o n e row. . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint Fig ur e 1. Out l ine of AD GA PP wor kflow u s e d t o pr io r itize ca ndidate gen es a nd va riants based on AD GWAS. 1) Left hand-side: We c on s i d er ed all th e ge ne s p osi tion ed within a 3Mbp in terva l c e ntr ed o n inde x GWA S S NP . Right hand-side: We c o ns i d ered va r i a nt s w ithin the in terv al ar o und top GWA S S NP d efin e d a s fo llow s: c on s i der a ll the SN P s within 1 M b p int erval aro und the ind ex SNP a nd find th e fu rth e st SN P (blu e c ircl e) in ei ther di rec ti on wi t h r 2 >= 0.2 i n 100 0 Geno me s EUR popul ati on – th e se v arian t s de fine t he bounda rie s o f the lo cu s in terval ( p ur p l e sh ading) wi t hi n whic h al l SNP s ar e co n s ide red . Fo r loc us in terval le ngth, w e found i t ran ge d f rom 28,1 33 bp to 91 5,373 b p, with me dian a t 228,670 bp. The number o f c andida te S NP s co nt a ine d w ithin a locu s int erva l varie d fr om 93 to 10,7 10, w it h a medi an o f 75 8 SNP s . 2) We a s sembled d i f f e re nt ty pe s of d ata s et s showi ng signi fic ant re sult s for ge n e s within the loc i ( left panel ), both g e ne s and S NP s ( middle panel ) and ju st SN P s ( right panel ). Tiles repr ese nt numb er o f da ta set s in e a ch c at egory a nd are fur the r col oure d ac cording t o subjec t i ve ev idenc e streng t h : r e d (highe s t): stati s tical te st s ba sed on ful l s u mmary sta ti s tic s , gray (middle) : l ookup s among s i gnific an t re s ul ts in e xpe rimen ta l studi e s, b lue (l ow ): predic tive ma chin e lear ning model s. 3) We s ummari sed t he ou tput o f f or ea ch e x periment /analy s i s in a set o f s t and ardi se d summary tab l e s . 4) We c alcu lat ed a f i nal score w hich allow e d ranki ng of a ll th e c on s i de red gen e s a nd SNP s fo r a giv en loc us, and p riori tis atio n of targe ts f or down str eam re se arch. Fig ur e 2. Sc ore by type of e viden c e for top 3 ranke d genes in th e 6 highlight e d l oci. Scores for top 3 ranke d gene s a t e ach locu s a re shown pa rtiti oned by categ o r y of ev ide nc e – he r e i nclud ing the top 10 c atego r ie s c on tr i bu t i ng the hi gh e st prop or ti on of t o ta l sco re a t the top 10 rank e d gene s f or all loc i. Or d e r o f loci corre s p o nd s to t he or d e r in Tab le 1. Fig ur e 3. Gene scores within the 3 Mbp inter v al of lea d SNP in the 6 highlighte d l oci. Top prioriti sed gen e marke d with a bl ac k squa re. A) loc u s 1q21.3 – b ; B) loc u s 2q12.1 ; C ) locus 5p 1 3.2; D ) loc u s 11p1 3; E) loc us 14q13 .2 ; F) loc u s 14q32. 32 . Fig ur e 4. Networ k visualisa tion of the func t iona l ter m s enric hed among locus to p 3 prioritis e d gene s in A DGAP P. The on t o logy catego ri e s a r e de picted a s blu e he xago n s , w ith t heir siz e line a r ly propor tional to -l o g 10 of adjus ted e nric h ment p -val ue . AD ge n es are d epic te d as pink r e ctang le s, wi t h t he int en sity o f the c ol our fil l pr o po r tiona l to gen e s c or e and thi ckne s s o f t h e gree n bo rder marki ng the g ene r ank at th e loc u s , with rank 1 the thic ke s t. Fig ur e 5. Highest -confidenc e i nteractions between locus top 3 prioritis ed genes in ADG A PP a nd other AD genes. AD gen e s priori ti sed ou t s i de o f A DGAP P ar e depic t ed a s oliv e r ecta ngle s. Fo r GW AS AD gene l ege nd, r efer t o Fig ur e 4. Edg es are col oure d ac cording to s ource o f evi dence for in tera ction : l ight blue - kn own interac tion s from cu r a ted da tab as e s, pink - expe r im e ntally det ermined k nown i n t e ra ction s, gr een - p r ed ict ed inter ac t i on s b a se d on ge ne n eig hbourh o od, red - predi c ted in ter ac tion s ba se d on ge ne fu sion s, d ark blue - pr edict ed int eracti on s ba sed on ge ne co-occ ur r e nce , black - co -e xpre s s ion , purple - prot ein homolo gy . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint DGE TWAS TWAS coloc methylation proteome PrixFixe regfm regulatory variant prediction TAD fine−mapping CTCF lncRNA mQTL promoter−enhancer active chromatin state caQTL ChIP−seq Conserved genomic regulatory blocks FAIRE−seq GWAS Catalog hQTL insulator piRNA splicing eQTL coloc promoter−enhancer eQTL with positive selection miRNA pQTL COGS eQTM hQTL mQTL Variant selection 248 x 344 x 1) 2) 3) 4) Final score top 3 * * a) b) c) + + * For each possible locus and gene/ variant combination, calculate final score: ~ proportional to result significance and effect ~ adjusted downwards based on n experiments, total SNP hits, total gene hits ~ adjusted upwards based on evidence weight and average number of studies & study types 3 Mbp 0 * Top GWAS SNP study id table id n experiments locus current SNP gene name FDR/p-value/posterior probability/score study type evidence weight total SNP hits total gene hits Unique study which data was sourced from Par�cular analysis/ experiment in the study Number of analyses/ experiments in the study dataset A gene�c locus harbouring AD GWAS hit rsid ID of a SNP within the GWAS gene�c locus Gene associated with current SNP or priori�zed by other methods, such as DGE A value indica�ng strength of evidence and/or magnitude of the effect Type of experiment (e.g. ChiP-Seq) or analysis (e.g. eQTL) Subjec�ve prior belief in analysis strength, from 1 (highest) to 3 (lowest) Total number of variants in a given locus found among significant hits in the given analysis/ experiment Total number of genes in a given locus found among significant hits in the given analysis/experiment Gene selection Top GWAS SNP * -0.5 Mbp + 0.5 Mbp r2 0.2 0.4 0.6 0.8 1.0 locus interval . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint 320 265 215 73 19 351 40 277 309 136 70 77 186 104 195 109 46 39 900 685 39 447 7 8 131 4 1 485 223 307 253 24 132 83 22 7 22 41 33 299 158 2 34 9 9 18 27 60 364 224 15 19 8 9 14 18 10 16 12 9 10 7 7 28 13 35 10 3 9 48 6 3 4 2 6 3 4 6 6 5 37 26 21 19 10 0 33 2 11 10 1 14 10 13 52 3 4 28 35 3 3 2 2 14q32.32 / rs7146581 CDC42BPB 14q32.32 / rs7146581 AMN 14q32.32 / rs7146581 TRAF3 14q13.2 / rs2038255 SRP54 14q13.2 / rs2038255 KIAA0391 14q13.2 / rs2038255 PPP2R3C 11p13 / rs2592555 COMMD9 11p13 / rs2592555 TRAF6 11p13 / rs2592555 PRR5L 5p13.2 / rs10214237 UGT3A2 5p13.2 / rs10214237 SPEF2 5p13.2 / rs10214237 IL7R 2q12.1 / rs6419573 IL1RL1 2q12.1 / rs6419573 IL18RAP 2q12.1 / rs6419573 IL18R1 1q21.3b / rs12730935 ADAR 1q21.3b / rs12730935 UBE2Q1 1q21.3b / rs12730935 IL6R coloc TWAS Hi−C eQTL DGE regfm PrixFixemQTL pQTL hQTL 0 100 200 300 400 500 600 700 800 900 score . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint A B C D E F . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint STAT3 establishment of skin barrier ATG16L1 HLA-DRAHLA-DRB1IRF1SOCS3 CLEC16AIL18RAP EGR2 RPTN STMN3 ETS1 NF-KB RELAIL4-mediated signaling HRNR Th1 and Th2 cell differentiation regulation of response to interferon-gamma IL7R IFNG JAK-STAT signaling pathway regulation of T cell differentiation IL2RA HLA-DQB1 interleukin 2 receptor complex interleukin-7-mediated signaling pathway interleukin 12 complex Selective expression of chemokine receptors during T-cell polarization cytokine-cytokine receptor interaction IL1RL1 INPP5D interleukin 23 complex IL18R1IL22 cellular response to cytokine stimulus JAK2 TRAF3 RBM17 IL6R DHX58 JAK1positive regulation of cytokine production TRAF6 HSPA1B . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint IL1RL1 TRAF3 CLEC16A TSLPR TSLP DEXIIL7R MAP2K7 HLA-DQB1 STAT3 TLR2 ADAR TRAF6 HLA-DRB1 IRF1 STAT6 HLA-DRA ETS1 IL4 IL22 TPD52 RAB5C IFNGR1 PAG1 IL6R SPRR3 RPTN LOR IL18RAP FLG DSG1 IL18R1 IVL IFNG IL2RA INPP5D SOCS3 FCER1G COMMD9 UBE2Q1 FCER1A . CC-BY-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.30.20240838doi: medRxiv preprint