The effect of variant interference on de novo assembly for viral deep sequencing

Christina J. Castro; Rachel L. Marine; Edward Ramos; Terry Fei Fan Ng

The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/815480 doi: bioRxiv preprint de Bruijn graph and overlap-layout-consensus assemblers produced better assemblies. 15 We speculate that 245 this "ensemble strategy" 15 may perform better because the multiple assemblers complement one another by 246 having different VI PID thresholds. Future assembly approaches could also consider resolving the VI problem 247 by possibly discriminating the major and minor variant reads first (perhaps by coverage or SNP analysis), and 248 then assembling major and minor variant reads separately. This study aimed to understand how variants affect assembly. As an initial investigation, many 259 confounding factors were simplified for experimentation. Simulated variants studied here only depicted 260 periodic mutations, set at regular intervals. However, in real viral data, SNPs are never evenly distributed 261 across the genome, with zones of divergence and similarity. 31, 32 Other important factors which influence 262 genome assembly include sequencing error rates, presence of repetitive regions, and coverage depth. We 263 limited our experiments to keep these factors constant in order to investigate the sole effect of VI. Through 264 this exploration, we demonstrated that reads from related genome variants adversely affect de novo 265 assembly. As NGS and de novo assembly have become essential for generating full-length viral genomes, 266 future studies should investigate the combined effects of the number and relative proportion of minor 267 variants, as well as additional assembly factors (e.g., error rates) to supplement this work. 268 269 All rights reserved. No reuse allowed without permission.