Tajima D test accurately forecasts Omicron / COVID-19 outbreak

20 21 On 26 November 2021, the World Health Organization designated the SARS-CoV-2 22 variant B.1.1.529, Omicron, a variant of concern. However, the phylogenetic and 23 evolutionary dynamics of this variant remain unclear. An analysis of the 131 24 Omicron variant sequences from November 9 to November 28, 2021 reveals that 25 variants have diverged into at least 6 major subgroups. 86.3% of the cases have an 26 insertion at amino acid 214 (INS214EPE) of the spike protein. Neutrality analysis of 27 DH (-2.814, p<0.001) and Zeng’s E (0.0583, p=1.0) tests suggested that directional 28 selection was the major driving force of Omicron variant evolution. The 29 synonymous (Dsyn) and nonsynonymous (Dnonsyn) polymorphisms of the Omicron 30 variant spike gene were estimated with Tajima’s D statistic to eliminate 31 homogenous effects. Both D ratio (Dnonsyn/Dsyn, 1.57) and DD (Dsyn-Dnonsyn, 0.63) 32 indicate that purifying selection operates at present. The low nucleotide diversity 33 (0.00008) and Tajima D value (-2.709, p<0.001) also confirms that Omicron variants 34 had already spread in human population for more than the 6 weeks than has been 35 reported. These results, along with our previous analysis of Delta and Lambda 36 variants, also supports the validity of the Tajima’s D test score, with a threshold 37 value as -2.50, as an accurate predictor of new COVID-19 outbreaks. 38 39 40 41 . CC-BY-NC-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 (which was not certified by peer review) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 3

42 43 On 26 November 2021, World Health Organization designated the SARS-CoV-2 44 variant B.1.1.529, Omicron, a variant of concern based on its unique mutations, and 45 unusual features suggesting that therapeutic monoclonal antibodies, such as 46 Regeneron, may be less effective against Omicron. Evidences that new mutations in 47 Omicron could have an impact on viral transmission or the severity of illness are 48 still under investigation (World Health Organization, 2021). Omicron variants were 49 first reported in the Gauteng province, South Africa on November 9, 2021, and 50 shortly it was detected in recent travelers to Belgium, Botswana, Canada, Hong 51 Kong, Austria, Australia, Portugal, Israel, United Kingdom, Netherland, and the 52 United States. One of the major concerns is that Omicron variants contain more than 53 30 mutations to the spike protein (A67V, Δ69-70, T95I, G142D/Δ143-145, 54 Δ211/L212I, ins214EPE, G339D, S371L, S373P, S375F, K417N, N440K, G446S, 55 S477N, T478K, E484A, Q493R, G496S, Q498R, N501Y, Y505H, T547K, D614G, 56 H655Y, N679K, P681H, N764K, D796Y, N856K, Q954H, N969K, L981F). Some of 57 these changes have been previously identified in Delta or Alpha variants and are 58 linked to heightened infectivity and the ability to evade infection-blocking 59 antibodies (Callaway, 2021). The likelihood of higher transmission rates has led 60 multiple countries to respond quickly to the Omicron variant. 61 62 The rapid increase in Omicron variant cases was found in the Gauteng province in 63 November, particularly in schools and among young people. Preliminary evidence 64 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 4 from genotyping tests suggests that Omicron may have been in circulation for quite 65 some time in South Africa (Callaway, 2021). To date, phylogenetic and evolutionary 66 status of Omicron variants are still not clear. Here we analyze the phylogenetic 67 relationship and selection pressure among the 131 available sequences of Omicron 68 variants. 69 70

71 72 131 complete SARS-CoV-2 genome sequences excluding low coverage from 73 November 9 to November 28, 2021 in 9 countries: Austria (N=1), Australia (N=1), 74 Belgium (N=1), Canada (N=1), Botswana (N=17), Hong Kong, China (N=2), Italy 75 (N=1), South Africa (N=105), and United Kingdom (UK, N=2) were collected from 76 the Global Initiative on Sharing All Influenza Data (GISAID) 77 (https://www.gisaid.org/). Sequence data in this study is available and deposited at 78 Figshare website (10.6084/m9.figshare.17105090). 79 80 FASTA files of viral sequences were downloaded and first aligned using MAFFT 7 81 software (Katoh and Standley, 2013, Kuraku et al, 2013). Phylogenetic relationships 82 between Omicron variants were analyzed using the neighbour-joining method and 83 Jukes-Cantor substitution model with bootstrap resampling number set as five. The 84 radial phylogenetic tree was generated by exporting the tree file in Newick format 85 by MAFFT. The FigTree software (version 1.4.2) was used to display the cladogram 86 (Rambaut, 2021). 87 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 5 88 The polymorphisms of the SARS-CoV-2 Omicron variants was analyzed based on the 89 site-frequency spectrum: (1) Tajima's D test, (2) normalized Wu and Fay’s DH test, 90 and (3) Zeng’s E test, using DNASP6 software with the bat SARS-CoV-2 WIV04 91 (GenBank MN996528.1) as the outgroup sequence (Rozas, et al., 2017). Statistical 92 significance of observed values of different tests was obtained by coalescent 93 simulation with intermediate recombination after 10,000 replicates. The probability 94 for each statistic was calculated as the frequency of replicates with a value lower 95 than the observed statistic (two-tailed test) by DNASP6 (Rozas, et al., 2017). 96 97 A modified Tajima’s D statistic was used to examine purifying selection based on 98 non-synonymous (Dnonsyn) versus synonymous sites (Dsyn) of SARS-CoV-2 genes was 99 calculated as previously described (Hahn et al, 2002, Hughes, et al., 2005, Yeh and 100 Contreras, 2021). The average number of pairwise synonymous differences (kS) 101 and the average number of nonsynonymous pairwise differences (kN), the number 102 of synonymous segregating sites (SS), and the number of nonsynonymous 103 segregating sites (SN) were computed with equation: (1) S*S=SS/a1, (2) S*N=SN/a1, 104 where a1 is defined as (Tajima, 1989, Hughes, et al., 2005). Dsyn was 105 defined as kS − S*S, divided by the standard error of that difference, and Dnonsyn was 106 defined as kN − S*N, divided by the standard error of that difference (Tajima, 1989, 107 Hughes, et al., 2005). 108 109

and Discussion 110 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 6 111 The insertion mutation Ins214EPE at the spike protein is present in some 112 Omicron variants 113 The phylogenetic tree identifies at least 6 major subgroups of 131 Omicron variant 114 sequences after rooting with an outgroup virus sequence of SARS-CoV-2 WIV04 115 from Wuhan, China (Figure 1). 113 Omicron cases (86.3%) contain an insertion of 116 nine nucleotides (GAGCCAGAA) between nucleotide 22204 and 22205 according to 117 WIV04 sequence (Figure 2). This generates an insertion of three amino acids 118 (Glutamic acid-Proline-Glutamic acid) (INS214EPE) in the N-terminal domain (NTD) 119 of the spike protein. 120 121 Resende et al. has reported that most ins214 motifs were rare in sequences of 122 different lineages of SARS-CoV-2 (A.2.4, B.1, B.1.1.7, B.1.177, B.1.2, B.1.214, and 123 B.1.429) with an insertion motif of 3 or 4 amino acids (AKKN, KLGB, AQER, AAG, 124 KFH, KRI, and TDR) (Resende, et al, 2021). Interestingly, the ins214 with four amino 125 acids (ins214GATP, ins214GATP, ins214GATS) were also present in bat SC2r-CoV 126 isolated in China (RmYN02), Thailand (RacCS203), and Japan (Rc-o319), 127 respectively. Although amino acid identity at ins214 vary among SARS-CoV-2 or 128 SC2r-CoV lineages, the insertion size (3 or 4 amino acids) is also conserved in 129 Omicron. It suggested that this region of NTD is susceptible to a mutation or 130 insertion. Whether or not INS214EPE affects spike protein function or immune 131 response requires further investigation. 132 133 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 7 Tajima’s D test is able to forecast new Omicron outbreaks 134 We have previously used neutrality tests to analyze the selection pressure of SARS-135 CoV-2 in the shipboard quarantine on the Diamond Princess (Yeh and Contreras, 136 2021a) and the influence of full vaccination coverage on Delta variants among 137 different countries (Yeh and Contreras, 2021b). We also proposed that Tajima’s D 138 test, a popular statistic in population and evolution genetics, can provide a 139 promising tool to forecast new COVID-19 outbreaks (Yeh and Contreras, 2021b). 140 141 Here the selection pressure of Omicron variants was first analyzed by multiple 142 neutrality tests. Within 131 Omicron variant sequences, 75 mutation sites were 143 identified with the nucleotide diversity (π, the average number of nucleotide 144 differences per site between two sequences) equal to 0.00008, which is significantly 145 lower than earlier outbreak of Delta variants in UK (0.0004-0.0006, N=376, March 146 26 to April 22, 2021), India (0.0004-0.0006, N=67, January 1 to March 11, 2021), or 147 Australia (0.0006, N=75, April 9 to May 6, 2021) (Yeh and Contreras, 2021b). Tajima 148 D test was calculated to compare π and total polymorphism (Tajima, 1989). Tajima’s 149 D values were negative and significantly deviated from zero (-2.709, P < 0.001) 150 among the whole genome sequences of Omicron variants. The negative values of 151 Tajima D were also detected in the ORF1ab (-2.617, P < 0.001) and the spike gene (-152 1.948, P < 0.005). This result indicates an excess of nucleotide variants of low 153 frequency (Table 1), and strong selection and/or demographic expansion was 154 operating in Omicron. 155 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 8 We have previously shown that Tajima D values decreased as SARS-CoV-2 Delta and 156 lambda variants spread in human populations. One to three weeks after Tajima D 157 fell below -2.50, Delta variant outbreaks emerged in India and the UK (Yeh and 158 Contreras, 2021b). Taken together, the low π and Tajima D values suggested that 159 Omicron variants have most likely spread within a population weeks before the 160 samples were collected. This data also agreed with our previous proposal that 161 Tajima’s D test is useful to forecast new COVID-19 outbreaks regardless of the 162 sample sizes of different variants (Yeh and Contreras, 2021b). 163 164 Purifying selection was operating in Omicron variants 165 Application of Tajima D is limited by the difficulty in distinguishing the influence of 166 both selection pressure and demographic expansion. To overcome this problem, we 167 included normalized DH test and Zeng’s E test in our analysis (Zeng et al., 2006, Yeh 168 and Contreras, 2021b). The normalized DH test is affected by directional selection 169 but insensitive to demographic expansion. Zeng’s E test is very sensitive to 170 population growth immediately after a sweep (Zeng et al., 2006). Genomic 171 polymorphisms of Omicron variants showed significantly negative DH values (-172 2.814, P0.1) (Table 1). The results were similar after confining our analysis to the spike 174 gene (DH, -9.014, p<0.001; Zeng’s E, 5.504, p<0.001) (Table 1), suggesting that 175 directional selection was the major driving force, without significant influence by 176 the demographic expansion. 177 178 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 9 We and others have shown the limitations to application of the inter-species 179 divergence, the dN/dS (w) test, for analyzing selection pressure of SARS-CoV-2 180 previously (Mugal, et al., 2014, Kang et al., 2021, Yeh and Contreras, 2021a, Yeh and 181 Contreras 2021b). Therefore, here we determined the purifying selection using a 182 modified Tajima’s D statistics instead of dN/dS (w) test. Under purifying selection, 183 the frequency distribution of non-synonymous polymorphisms is negatively skewed 184 relative to the distribution of synonymous polymorphisms. Therefore, it takes more 185 negative values for non-synonymous (Dnonsyn) than for synonymous sites (Dsyn) of a 186 given gene (Hahn et al, 2002, Hughes, et al., 2005, Yeh and Contreras, 2021b). One 187 of the major advantages of Dnonsyn and Dsyn analysis is that it is independent of 188 sample size, which allows us to compare Dnonsyn and Dsyn values among different data 189 sets (Hughes et al., 2008). The same excess of low-frequency alleles in non-190 synonymous polymorphism was also shown by the Dnonsyn values of the spike (-191 1.771, p<0.001) and N gene (-1.943, , p<0.005) (Table 1). We conclude that 192 purifying selection led to constraints on the neutral mutations at non-193 synonymous sites of the spike gene of Omicron variants. 194 Negative values of Tajima D could be caused by a bottleneck event rather than 195 selection, but the bottleneck effect should affect all types of polymorphism equally 196 (Tajima, 1989). Hahn et al. have reported that Dnonsyn is disproportionally lower than 197 Dsyn, because non-synonymous and synonymous mutations are affected unequally 198 by purifying selection (Hahn et al., 2002). The value of DD (Dsyn-Dnonsyn) increased as 199 purifying selection became stronger, and DD can eliminate the homogenous effects 200 (demographic expansion, selective sweep etc.). DD values of the spike and N gene of 201 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 10 SARS-CoV-2 Omicron variants are positive (0.643 and 0.377, p<0.005) (Table 1). 202 Austin Hughes showed that the standard error terms in both equations of Dnonsyn 203 and Dsyn cancel out the sample size effect when D ratio (Dnonsyn/Dsyn) was applied to 204 the data analysis (Hughes, 2008). D ratio values of the spike and N gene were 205 significantly more than 1 (1.57 and 1.24, Table 1), suggesting that purifying 206 selection pressure of Omicron spike and N gene was operating. The values of 207 DD and D ratio of ORF1ab (-0.05 and 0.980, p<0.001) were insignificant, consistent 208 with the idea that the spike and N gene have been under more selective constraint 209 than ORF1ab (Chaw et al., 2020). So far, no evidence has shown that vaccination or 210 other mitigation procedures causes positive selection of Omicron variants. 211 212 Perspectives 213 Previously we found that one to three weeks after the D value fell below -2.50, Delta 214 outbreaks emerged in India and UK, and the lambda variant outbreak in south 215 America (Yeh and Contreras, 2021b). This led to our proposal that Tajima D test 216 with a cut-off threshold value as -2.50, can predict new SARS-CoV-2 outbreak (Yeh 217 and Contreras, 2021b). Despites of the small sequence sample size (131 samples) in 218 this study, we detect a strong negative value of Tajima D in Omicron variants. This 219 finding also confirmed that the Omicron outbreak had emerged sometime before the 220 current breakout as the Tajima D test is sensitive to detect it and an efficient 221 predictor of future outbreaks. This study demonstrates that rapid genomic sequence 222 surveillance is essential, and Tajima D’ tests should be included to forecast future 223 outbreaks in different geographic populations. 224 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 11 225 Disclosure statement 226 No potential conflict of interest was reported by the author(s). 227 228 Author contributions 229 All authors contributed to study concept, rationale, and initial manuscript drafts, 230 interpretation of data, and final manuscript preparation. All authors have read and 231 approved the final version of the manuscript. 232 233 Funding 234 No funding 235 236 Ethical approval 237 None declared. 238 239

240 241 Callaway, E. (2021) Heavily mutated Omicron variant puts scientists on alert. 242 Nature. https://doi.org/10.1038/d41586-021-03552-w 243 Chaw, S.M., Tai, J.H., Chen, S.L., Hsieh, C.H., Chang, S.Y., Yeh, S.H., Yang, W.S., Chen, P.J., 244 Wang, H.Y. (2020) The origin and underlying driving forces of the SARS-CoV-2 245 outbreak. J Biomed Sci., 27, 73. 246 Katoh, K, Standley, D.M. (2013) MAFFT multiple sequence alignment software 247 version 7: improvements in performance and usability. Mol Biol Evol. 30, 772–80. 248 Kuraku S, Zmasek CM, Nishimura O, Katoh K. (2013) aLeaves facilitates on-demand 249 exploration of metazoan gene family trees on MAFFT sequence alignment server 250 with enhanced interactivity. Nucleic Acids Res.,41, W22-8. 251 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 12 Hahn, M.W., Rausher, M.D., Cunningham, C.W. (2002) Distinguishing between 252 selection and population expansion in an experimental lineage of bacteriophage T7. 253 Genetics, 161, 11-20. 254 Hughes, A.L., (2005) Evidence for Abundant Slightly Deleterious Polymorphisms in 255 Bacterial Populations. Genetics, 169, 533-538. 256 Hughes, A.L., Friedman, R., Rivailler, P., French, J.O., (2008) Synonymous and 257 nonsynonymous polymorphisms versus divergences in bacterial genomes. Mol Biol 258 Evol. ,25, 2199–2209. 259 Kang, L., He, G., Sharp, A.K., Wang, X., Brown, A.M., Michalak. P., Weger-Lucarelli, J. 260 (2021) A selective sweep in the Spike gene has driven SARS-CoV-2 human 261 adaptation. Cell, 184, 4392-4400. 262 Mugal, C.F., Wolf, J.B., Kaj, I. (2013) Why time matters: codon evolution and the 263 temporal dynamics of dN/dS. Mol Biol Evol., 31,212–231. 264 Resende, P. C., Naveca, F. C., Lins, R. D. et al. (2021) The ongoing evolution of 265 variants of concern and interest of SARS-CoV-2 in Brazil revealed by convergent 266 indels in the amino (N)-terminal domain of the Spike protein. DOI: 267 10.1101/2021.03.19.21253946 268 Rozas, J., Ferrer-Mata, A., Sánchez-DelBarrio, J.C., Guirao-Rico, S., Librado, P., Ramos-269 Onsins, S.E., Sánchez-Gracia, A. (2017) DnaSP 6: DNA Sequence Polymorphism 270 Analysis of Large Data Sets. Mol Biol Evol., 34, 3299–3302. 271 Rambaut A. FigTree, version 1.4.2. San Fransisco: GitHub, Inc.; 2021. Available from: 272 https://github.com/rambaut/figtree/releases 273 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 13 Tajima, F. (1989) Statistical method for testing the neutral mutation hypothesis by 274 DNA polymorphism. Genetics, 123, 585–595. 275 World Health Organization. (2021) Update on Omicron. 276 https://www.who.int/news/item/28-11-2021-update-on-omicron. 277 Yeh, T.Y., Contreras, G.P. (2020) Emerging viral mutants in Australia suggest RNA 278 recombination event in the SARS-CoV-2 genome. Med J Aust ,213, 44-44.e1. 279 Yeh, T.Y., Contreras, G.P. (2021a) Viral transmission and evolution dynamics of 280 SARS-CoV-2 in shipboard quarantine. Bull World Health Organ., 99, 486–495. 281 Yeh, T.Y., Contreras, G.P. (2021b) Full vaccination is imperative to suppress SARS-282 CoV-2 Delta variant mutation frequency. MedRxiv, 283 https://doi.org/10.1101/2021.08.08.21261768 284 Zeng, K., Fu, Y.X., Shi, S., Wu, C.I. (2006) Statistical tests for detecting positive 285 selection by utilizing high-frequency variants. Genetics, 174, 1431-1439. 286 287 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 14 Figure Legend 288 Figure 1. Rooted phylogenetic tree of SARS-CoV-2 genomes of Omicron variants, 289 November 9 to November 28, 2021. Alignments of viral sequences were generated 290 using MAFFT, and the phylogenetic tree was visualized using FigTree. Rooting was 291 done by introducing SARS-CoV-2 WIV04 (red line, accession number MN996528) as 292 an outgroup virus. 293 Figure 2. Alignment of Omicron variant with and without insertion at the amino acid 294 214 of the spike protein. 295 296 297 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 15 Figure 1. 298 299 300 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 16 Figure 2. 301 302 303 304 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint 17 Table 1. Neutrality analysis of SARS-CoV-2 Omicron variants of the full genome and 305 the viral genes. The statistical significance was estimated using 10,000 coalescent 306 simulations in DNASP6. Not significant values (P>0.05) are indicated in italic. NS8 is 307 excluded in this table because no mutations were detected. (N.D., not determined; 308 N.A., not applicable). 309 310 Whole genome ORF1ab Spike NS3 E M NS6 NS7a NS7b N DH -2.814 -0.152 -9.014 -1.875 -5.264 -7.552 0.043 0.0588 0.204 -3.016 Zeng's E 0.053 -2.138 5.504 0.149 3.272 4.603 -0.725 -0.996 -0.598 0.716 Tajima's D -2.709 -2.617 -1.948 -1.706 -0.735 -1.281 -0.998 -1.342 -0.65 -2.237 DNonsyn N.D. -2.41 -1.771 -1.342 N.A. N.A. N.A. N.A. N.A. -1.947 Dsyn N.D. -2.46 -1.128 -1.34 N.A. N.A. -0.998 -1.342 -0.65 -1.57 D ratio (Dnonsyn/Dsyn) N.D. 0.980 1.570 1.001 N.A. N.A. N.A. N.A. N.A. 1.240 DD (Dsyn-Dnonsyn) N.D. -0.05 0.643 0.002 N.A. N.A. N.A. N.A. N.A. 0.377 . CC-BY-NC-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) The copyright holder for this preprint this version posted December 2, 2021. ; https://doi.org/10.1101/2021.12.02.21267185doi: medRxiv preprint