Larva in the loop, a closed loop machine interface system forDanio reriolarvae

Larva in the loop, a closed loop machine interface system for Danio rerio larvae John Jut oy 1 [0000−0003−4038−5112] and E ri ca Jung 1 [0000−0002−1426−7537] 1 Department of Mechanical and In dus tria l Engine ering , Un iver sity of Ill in ois at Chicago , Chicago Il 606 07, USA Abs tr a ct. The opt okinet ic r e spons e (OKR) in Z eb r af ish ( D ani o R er io) had be en char act er iz ed f or it s r obus t r e spons e t o vi sua l s ti muli. Expanding on the se w o rk s, w e developed a n ovel cl osed lo op c on tr ol sche ma t o d rive a r obo t ut ilizing the OKR o f Z ebr afish l a r vae. Ou r s y s t em k e eps the b o dy of a la r v a cons t r ained v ia a n ovel ag ar os e mo ld h olde r tha t all o ws f o r ey e m ovemen t and v i si on. The la r v a is th en put under a mic r os c ope c am er a and pr oc ess ed t hr ough c omput e r visi on t o t r a ck i ts ey e s via elli pse fi t ting. R ela t iv e ey e angle da t a i s then pa r sed th r ough an al go ri thm and used t o send m ove men t signa ls t o a r obo t on a l ined t r a ck. Simu lt ane ously , th e r obo t re t ur n s it s rel at i ve p os i ti o n wi t h res p e ct to t h e l i n e a n d co nve r t s t h at i n f or m a t io n in t o an OKR s t imula tion ani ma ti on which is displa y ed on an L CD s cr e en in the ven t r a l plane of the l a r va, thus cl osing the l oop. Thr o ugh this w ork we sho w the c apabili ty o f la r vae OKR t o k e ep a r ob ot on a line ar t r ack aft e r an initi al oblique en tr an ce t o th e line. Thi s w o rk di spla y s the pot en tial of ou r s y s t em and h ow it can pa ve the w a y t o a Z ebr afish B r a in-Ma chine In t e rf a ce. K eyw or d s: Z eb r afish La r vae · Opt okine tic R es ponse · F e edback C on tr ol 1 Introduction T o under s t and and pe rhaps con t r o l the hum an br ain, i t i s k ey t o ha ve t o ols tha t c an monit o r and e lic it r e spons es in the b r a in. Al though gr ea t s tr ides ha ve b een mad e in mapping and in t erf acing wi th the hum an br ain, ther e is me rit t o s tudying o r g anis ms w i t h l e ss c om p l ex bra i n syste ms s u c h a s C. El ega ns, D r os op h il a , and Da nio R erio 1 . Animal b r ain mod els can help in under s t a nding neur opa tho logy 2 , i mpr oving br ain mapping 3 , o r devel oping bet t er me thod ol o gies f or bi o- mach ine in t erf ac es, a ll of which we hope t o addr ess wi th our s y s t em a nd its fu tur e it e r a t ions. Z ebr afish la r va e ar e ide al m odel o r g anis ms due t o the ir f as t r epr oduct ion r a t e, e xt ernal devel opm en t, and t r anslu cency 4 . Z ebr afish l a r vae tr anslu cency c ombin ed with g ene tic engine ering t o ols and opti cs (opt og enet ic s), all ow f or non-in vasive neur al a ctivity me asur e men t and s ti mula t io n 5 . The w ork pr es en t ed her e des c ribes the pr ogr ess t ow ar ds a neur al m onit o ring an d eli cit a t ion s y s t em f o r Z ebr afi sh la r vae. Thus f ar , we ha ve b een abl e t o d evel op a v i s ual s ti mula ti on s y s t e m tha t w ork s in c onjuncti on w ith a la r va e h older . W e d is pla y the cap abili ties of the devel oped Z ebr afish Ma chine In t e rf a ce which util iz e s opt okin et ic r esp onse in b oth open and was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 13, 2024. ; https://doi.org/10.1101/2024.07.12.603215doi: bioRxiv preprint 2 J. Jutoy , E. Jung clo sed lo op m anner . W e w ill discu ss h ow the w ork pr es en t ed is a s t eppings t on e t ow a r ds our g oal of a Z eb r af ish B r ain- Ma chine In t erf ac e. 1.1 Opt okine ti c R esp onse T o valida t e ou r s y s t em, w e ai med t o r epr o duce wha t ha s been w idely d one wi th Z ebr afish La r vae whi ch is t o de m ons t r a t e i ts opt okine tic r espons e thr ough v isua l st i m u l a t io n 6 . The Opt okine tic r e spons e ( OKR) is one of th e g az e s t abiliz a t ion me chanis ms which m oves the ey e such tha t a f ea tur e of in t er es t is k ep t in f ocus. The Z ebr afish la r va e OKR has been us ed f o r v isual s cr eening o f mut an ts due t o its r obus t and r el iable r espon se 6 and has been used t o quan tify Z eb r afish visua l a cuity 7 . The w o rk pr esen t ed he r e capit aliz es on thi s r obus t r e sponse by in tr oducing it in a clo sed f eedba ck s y s t em. W e p osi t tha t the OKR r e spons e of the Z eb r afish can main t ain a se t-po in t wi th pr ope r pa r am et e r t uning of O KR s ti mula t ion. 1.2 Novel L a r vae Fix a ti on Consis t en t and ac cur a t e ey e t r acking r equi r e s la r va he ad fix a ti on w ithou t obs t ruc ting ey e m ovemen t o r visi on. Ag a r o se emb edding 6,8 beca me the s t andar d in Z eb r a fish la r vae fix a ti on due t o its t r anspa r en cy and non-t o x ic a t tr ibut es. Al though ag ar os e embedding h as be co me the s t andar d, i t is g ener ally a t im e-c onsu ming p r o ce ss. W e in tr oduc e in this w o rk a nove l and m or e ef fi cien t m ethod f or la r va e fix a ti on inspi r ed by Coppe r e t. a l 9 . 2 Methods A m odula r appr oa ch w as t ak en t o devel op the Z eb r afish M achin e In t e rf a ce s o tha t modul es can be e asily added o r imp r oved in futur e w o rk. B ri ef des cr ipti ons a r e pr ovid ed f or individual modul es al ong wi th t he la r v ae c ar e m e thod ol ogies. The open-l o op s y s t e m, O KR visua l s t imula t ion, and ey e t r a cking wer e us ed in c onjuncti on t o ve rify tha t the s y s t em cou ld p r ope rly eli ci t OKR in l a r vae. The cl osed-l o op s y s t em is s imply the open-l oop s y s t em w ith the addit ion of th e r ob ot ca r modul e. 2 . 1 L ar vae F i xat io n A r es in 3D p rin t ed p osit ive mo ld w a s cr ea t ed wi th f ea tur es of a Z eb r af ish l a r va as seen in figur e 1 c. The di mens ion s w er e such tha t it w ould cons tri ct la r va m ovem en t while h a ving en ough spa ce f o r ey e m ov e men t. Cr ea t ion of the d evic e cons is t ed of hea ting up ag a r o se g el in t o it s liquid f or m and placing it in t o a pe tri di sh with th e p osi ti ve m old. The ag ar ose so lidifi es in t o a tr ansluc en t gel in which the p osi tive mo ld c an be c ar efully r em oved. N eg a tive spa ce lef t by the p osi tive m old i s then fi lled with a l a r va th r ough a ha ir l oop dev i ce 10 . was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 13, 2024. ; https://doi.org/10.1101/2024.07.12.603215doi: bioRxiv preprint 3 Figure 1 S y s t em Setup of Z ebrafish Machine Int erf ace including s ys t em f low chart, renderings of hardware c om pone nts, and image pr ocessing of eye angle. (a) S y s t em logic f lowchart with key paramet ers (ω= cons t ant an gular speed of s ti muli animation, θ= minor axis angle of f itt ed ellipse on larva eye with re spect t o positive horiz o nt al axis, x = di s t ance betwe en ce nt er of R.Pi camer a and ce nt er of yellow line) highl ight ed and colored . (b) R endering of Z ebrafi sh s y st em (c) Closeup of Z ebrafish larva holder . (d) Animation of rot ating black and w hit e arcs at cons t ant angular speed ω provided by L CD screen t o elicit o pt okinetic response. (e) Z oomed in image of greyscale larva v ideo frame along with region of int eres t (R. O .I.) highlight ed. (f ) R. O .I. passed throug h Gaussian filt er t o iso lat e lar va eye. (g) E ll i pse fitt ed t o (f ) overlayed int o (e) t o e xtract θ. (h) R endering of R.Pi s y s t em. (i) Z oomed in rendering of R.Pi car t o disp lay the horiz ont al dis t ance betwee n ca mera and line. 2.2 OKR Vi sual St imula ti on A m odule in the s y s t em s oftw a r e w as cr ea t e d t o pr ovide a r ot a t ing g r a ting ani ma ti on using the Op enCV pyth on libr ary . S ever al pa r a met er s c an b e eas ily m odifi ed by th e s y s t em inc luding gr a ting r ot a t ion speed ω , number of gr a tings, g r a ting ar c l eng th, and gr a ting r adius. Suf fici en t c on tr as t f o r the ey e t r a cking m odule w a s pr ovided by including a blank whit e ci r cl e di r ec tly unde r t he fish. The par am et e r s chos en f or t es ting wer e i den tified m anually ba sed on visual inspect ion of OKR fr o m the la r va e. Thes e p ar am et e r s w er e f ound t o be: 5 g r a tings, 50% gr a ting sp acing (ie: half of th e g r a ting is black, and half is whit e ). 2.3 E y e T r a cking The fish ey e t r a cking module c on vert s i ma ges fr om the mic r os c ope c am er a v id eo s tr ea m in t o ey e angles of the Z eb r afish ey e of in t er es t. Using the Open CV lib r a ry of python, r aw im age s ar e c on vert ed t o gr ey s c ale, blur r ed, then thr e shold filt er ed t o a was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 13, 2024. ; https://doi.org/10.1101/2024.07.12.603215doi: bioRxiv preprint 4 J. Jutoy , E. Jung value f ound suit abl e f o r i so la ti on of the Z e br af ish ey e. C on t our ing is done on the isola t ed ey e and an e llip se is then fit t ed t o it. The m ajo r a xis of the e llip se is then used as the angl e of the ey e. 2.4 R obo t C ar A r a spber ry pi c ar wi th tw o wh eel ed m ot or s in the fr on t and a f r ee m oving whee l in the ba ck w as ch osen as the vehi cle f or the Z ebr afish t o c on tr ol. The modif ied chas sis w as made t o ac c o mm oda t e a r aspbe rry pi c ame r a. Co m munica ti on t o the r aspbe rry pi c a r f r o m the ma in w o rk s t a ti on w as done t hr ough th e P a r a mik o lib r a ry . The r ob ot ca r w as t r a ck ed th r ough a L og it ech Bri o w eb ca m and pr o cess ed thr ough OpenCV t o e xt r act i ts es ti ma t ed p osi tion. 2.5 Z ebr afish La r vae La r va e we r e br ed fr o m wild-typ e adult i n a dedi ca t ed fish r o om, t e mpe r a tur e c on tr oll ed t o 26 /i1C , with a 14-h ligh t an d 10-h darknes s cy cle. The la r va e a r e har ves t ed and k ep t in the fi sh r o o m un til 4- 5 da y s pos t f e rti liz a ti on (dpf ) fr o m whi ch they ar e br ough t t o the t es t ing f ac ility a d a y prio r f or e xper imen ts. La r vae a r e k ept in an incuba ti on cha mbe r s et a t 27 /i1C outsid e o f e xper im en t al tri als. 3 Results 3.1 V eri fi ca ti on of Opt okin et ic R espon se T o ve rify tha t OKR is displ a y ed by Z eb r afish la r vae, a si mple t es t w ith a du r a ti on of 140 se c onds w as c r ea t ed t o analy z e ey e r esp onse t o the devel oped OKR s ti muli. The fir s t and las t 10 se c onds of the t es t hide th e individual a r c f ea tur e s of the s t imuli t o se r ve as a ba sel ine of no r e spons e and i s r e f e rr ed t o as ω = 0 . The s ti muli ar e displa y ed a t ω = 5 , 15, 30 , 45, 60 degr ees / se c in 20 se c ond ti me in t e r val s b et we en 20 - 120 r esp ec tively . W ithin ea ch 20 s e cond ti m e bin, a di r ec ti on swit ch is made. Analy sis of t ot al ey e angle r e spons e ( figur e 3) w as done t o di spla y tha t the OKR r esp onse w as r ela t ed t o th e visu al s ti muli of t he s y s t em. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 13, 2024. ; https://doi.org/10.1101/2024.07.12.603215doi: bioRxiv preprint 5 Figure 2 Right eye angle response f or a single Z ebr afish la rva at vary ing direction and ω as seen in figure 1 -d. E ye angles are in t erms of degrees while ω are in t erms of degrees/ sec. Not e that eye an gle slope direction agrees with s timul i ω direction within times that the s timuli was given: 20 -12 0 sec. 3.2 La r va e in the L o op Line F o ll owing Once opt okinet ic r esp onse w as iden tified in a fish, tr ial s w er e conduct ed t o se e if the l ar va e ca n m ai nta i n a r o bo t c ar wi t h i n a l in e at d i ffe r e nt ω as seen in figu r e 5. A r epr e sen t a tive se t of t ri als f r o m a singl e la r va drive r can be se en in figur e 6. E a ch tri al c onsis t ed of s t a rt ing the r ob ot c ar f a cing th e y ell ow l ine a t a sl igh t r and om obl ique angle. The ca r w a s tr ack ed and dr iven by th e la r vae f or a dur a ti on of 45 se c onds. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 13, 2024. ; https://doi.org/10.1101/2024.07.12.603215doi: bioRxiv preprint 6 J. Jutoy , E. Jung Figure 3 Absolut e cumulative eye angle of left and right eye co mpare d t o s timuli speed ω was done t o display that larvae response i s due t o our visual s t imuli s y s t em. ω = 0 represents no visual s t imuli present. Figure 4 Larvae in the loop line f ollowing real time view . A larva with it s left eye being tracked superimposed on t op of the visual s timul i it s receiv ing (left) and the robot car bein g co ntrolled by the Z ebrafish larva ( right). The change in left eye angle is checked with a threshold in order t o send a move signal (left if |θ| threshold, and f orward if within threshold) A camera in fron t of the robo t car sends the location of where it las t saw the yel low line t o the vi sual s timuli s y st em adjus ting direction/display respectivel y in order t o di rect the larva’ s eye t o back t o the direction of the yellow l i ne. Not e that ω is kept cons t an t throu gho ut the driving trial. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 13, 2024. ; https://doi.org/10.1101/2024.07.12.603215doi: bioRxiv preprint 7 Figure 5 P athway trac es of the Z ebrafish dr iven car at diff erent ω. The t op number on the plots represents ω in degrees / sec and each plot repres ents a trial with a duration of 45 seconds. Axis i s in t erms of pix el s. The robot car is s t art ed in itiall y at the bott o m right. 4 Discussion 4.1 Novel L a r vae Fix a ti on M eth od The la r vae f ix a ti on m eth od in tr oduced in t his w ork su cc eeded in k eeping l a r vae in place t o run cu mula t ive v i sual e xpe ri men ts up t o s ever al m inut es l ong. Visua l e xperi men ts can el ici t s t ar tle r esp onses , whi ch we r e ob se r ved in s o me t ri als, s ti ll ou r device w a s able t o k e ep th e la r v a e f ix ed and within the r eg ion of in t er es t. F or the us e ca se in this pape r , this devi ce m ade t ria l s signif i can tly m or e ef fic ien t as m ino r dec oupling of a l a r vae and the d ev ic e can be t o le r a t ed but m a y be inappr opr ia t e f o r othe r c ase s wh er e sligh t m ove men ts ar e no t t ole r abl e. 4.2 Mech anis m of Visua l Sti muli The m echani sm behind opt okine tic s ti muli p r oje ct ed fr om bel o w th e l a r vae w a s n ot the f ocus of this pr ojec t. W e p osi t tha t it m a y be due t o th e fish r e sponding t o the visual s ti muli be l ow i t as in opt o m ot o r r esp onse s etup s. Alt e rna tiv ely , i t m a y b e due t o the r ef r act ion of ligh t a cr os s var ious in t e rf ace mediu ms tha t ar e c ausing a f r on t al- la t er al pr oje cti on t o the l a r vae. I t m a y al so be a co mbina ti on of the tw o men ti oned theo ri es. R eg ar dle ss, opt okine ti c r espon se i s cl ea rly pr esen t fr o m the pr ojec ti on of the s ti muli b el ow the l a r vae as seen in 2 as the ey e angle r esp onse agr ees with the s timuli di r e cti on and is sp or adic or n ot p r esen t when th e s ti muli is e ithe r no t displa y ed or paused. 4.3 Move men t Th r esh olds was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 13, 2024. ; https://doi.org/10.1101/2024.07.12.603215doi: bioRxiv preprint 8 J. Jutoy , E. Jung Indiv idual la r va e displa y va ria ti on in OK R sensitivity . As such, the m ove men t thr esh old uti liz ed f o r an OKR s ensi tive la r va w as not suf fici en t f o r a l a r va w ith le ss OKR sens itivi ty . Futu r e w o rk shou ld iden tify t his thr esho ld and c o mpar e th e dr iving of la r vae wi th va rying OKR sen sitivity . V a ria t io n in OKR sen sitivity m a y be due t o the belo w ven tr al pr oje cti on of v isua l s ti muli a s discus sed ab ove. An othe r e xplana ti on ma y be due t o sligh t mis align men ts b et we e n la r va, f ix a ti on dev i ce , and mic r os c ope. Since the r e is a r ang e of t ole r an ces depend ing on l a r vae si z e and h y dr a ti on of th e a ga r o se , s l i g ht d orsa l mi s al i g n m e nt s w ere no t ic e d i n so me tr i al s . Howe ve r , t h ere i s als o p ossibi lity tha t thes e OKR sens itiviti es o bse r ved th r ough this s y s t em ma y be an aspec t of the Z ebr afish la r vae n ot in v es tig a t e d bef or e. 4.4 Futur e W o rk W e e mphas iz e tha t the w o rk pr esen t ed her e is a fundam en t al s t ep t ow a r ds developing a Z ebr afi sh B r ain- Machin e In t e r f ace. T o map the br ain of an o r g anis m thr ough high th r oughput inputs, w e r equir e a w a y t o si mult an eous ly s ti mula t e and r e co r d the la r vae. In additi on t o this, we mus t k eep the la r vae in pla ce wi thou t in t erf e ring w ith the input s ti muli and output r esp onse s. Alth ough the cur r en t output r esp onses we util iz e in thi s w ork a r e no t e xpl icit ly neu r a l signals , we de m ons tr a t e the pot en tial of ou r s y s t em t o vi sually s t imula t e th e la r v a e which we ai m t o map t o neur al a ctivity . F u r t he r c h arac ter i zati o n b e s i d es ω wi ll be e xplo r ed in the futur e. V isual par a m et e r s such as the spa cing be t ween gr a tings, numb er of g r a tings, and dis t ance of g r a t ings f r o m the la r vae m a y af f e ct the r es ponse. As men ti oned in the OKR sec ti on of m eth odo logy , the visua l pa r a met er s w er e tuned m anually bas ed on r ea ctiv ity o f the spec ific l a r vae. Whe ther th er e e x is t r e la t ionships b et we en these pa r a m et e r s and the l a r vae wil l be ana lyz ed in d et a il in a futur e paper . W ork is planned t o fu rthe r e xpl or e the n ovel la r va e f ix a ti on me thod des crib ed in this sh or t paper . A lth ough the over ar ching c oncep t w a s pr ov ided, in depth dis cuss ion w as n ot included h er e as cu rr en t w o rk i s ong oing t o cha r a ct e riz e fix a t ion pa r a met er s. 5 Conclusion This w ork displ a y s tha t Z eb r afish l a r vae c an be uti liz ed as con t r ol le r s t o s impl e s y s t ems l ik e lin e f oll ow ing r obo ts. Fur the r e xperi men t a ti on is r equi r ed t o e lucida t e s y s t em r ela t ionship s, h o weve r , with c o ar s e par a m et e r s et tings ( i.e. arbi t r ary s et ting of thr esho lds) pr om ising r e sults c an b e g ain ed. Furth er m or e, we b ri efly sh ow cas ed our novel , f as t, and ef fici en t t e chnique f o r f ixing a la r va in pl ac e w ith the tr ade- of f tha t s t a rt led l a r vae ma y ge t s ligh tly mis aligne d. Acknowledgments. This work w as suppo rt ed by t he Nationa l Science Foundatio n [Neu r on -to- Neur o n Int erf ace : Optical ly Connec t ed Neu r o ns B etwe en the Br ains of T wo Z eb r afish. Gr ant#: 2309 589 ]. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 13, 2024. ; https://doi.org/10.1101/2024.07.12.603215doi: bioRxiv preprint 9 Disclosur e of Int er es ts. The autho r s ha ve no comp eting int er es ts to declar e that ar e r e lev ant to the cont ent of this a rticle. References 1. St ew a rt , A. M. & K alu ef f , A. V . D eveloping b e t t er and m or e valid ani mal m odels of br ain d is or de r s. Be ha v . B r a in Res . 276, 28–31 (2015). 2. Raz ali, K. et a l. T he P r om i s e o f t h e Ze br af i s h M o de l f o r P ar k i n s o n’ s D i s ea s e: T oda y ’ s Sci enc e and T o mo r r o w ’ s T r ea tm en t. F ron t. G en et. 12, (2021). 3. Randlet t, O . et a l. Who le-b r ain a ctivity m apping on t o a z ebr afi sh b r a in a t las. Na t . Me th ods 12, 1039–1046 (2015). 4. Bes t, J. D . & Alde rt on, W . K. Z ebr afi sh: An in viv o m odel f or the s tudy of neur ol ogi c al di sea ses. Ne uro ps ychi atr . Dis. T reat. 4 , 567–576 (2008). 5. Sim mi ch, J., St a yk ov, E. & S c ot t, E. Chapt e r 8 - Z ebr afi sh as an appea ling mode l f o r opt o gen eti c s tudi es. in Pro gre ss in Brain R e s earch (eds. K nöp f el, T . & B oy den, E. S.) vol. 196 145–162 (Els ev ie r , 2012). 6. Br o ck erh of f , S. E. et a l. A beha vi o r al s cr een f or is ola ting z ebr afi sh mut an ts wi th visual s y s t em d ef ec ts. Proc. Nat l. Ac ad. Sci. 92 , 10545–10549 (1995). 7. Cam er on, D . J. et al. The Opt okine ti c R esp onse as a Quan tit a t ive Me asur e of Visual Acui ty in Z ebr afi sh. J . V i s . E x p . J oV E 50 832 (2013) doi:10.3791/50832. 8. Huang , M. Y .- Y . & Neuhaus s, S. The opt oki neti c r e spons e in z eb r af ish and it s appli ca ti ons. F r on t. Bio sci. J. Vir tua l Libr . 13, 1899–916 (2008). 9. Copper , J. E. et a l . C omp ar a tive analy si s of fix a tion and e mbedding t e chniques f or opti mi z ed his t ol ogi cal p r epa r a t ion of z ebr af ish. Comp. Bioc hem. Ph y sio l. P art C T o xico l. P harm aco l. 208, 38–46 (2018). was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 13, 2024. ; https://doi.org/10.1101/2024.07.12.603215doi: bioRxiv preprint 10 J. Jutoy , E. Jung 10. Benar d, E. et a l . Inf ec ti on of Z eb r afish E mbry os wi th In t r a cellu la r Ba ct e ria l P a tho gens. J. Vi s. Ex p. J oVE 61, (2012). was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 13, 2024. ; https://doi.org/10.1101/2024.07.12.603215doi: bioRxiv preprint