A binning tool to reconstruct viral haplotypes from assembled contigs

Jiao Chen; Jiayu Shang; Jianrong Wang; Yanni Sun

With or without reference genomes, genome-scale haplotype reconstruction in quasispecies remains a computationally challenging problem. High similarity between haplotypes in the same quasispecies and the heterogeneous sequencing depth along the viral genomes present barriers to adoption of existing assembly programs. A recently published comparison showed that none of the tested haplotype reconstruction tools were able to successfully reconstruct the five known strains for a mock HIV quasispecies (Jayasundara et al., 2014) . We had the same observations when comparing several popular metagenomic assembly tools and haplotype assembly tools such as IDBA-UD (Peng et al., 2012) , IVA (Hunt et al., 2015) , SAVAGE (Baaijens et al., 2017) , MLEHaplo (Malhotra et al., 2015) on the same data set (Chen et al., 2018) . Many methods output a set of contigs with various sizes that are much shorter than the genomes. With these outputted contigs from assembly programs, it still remains to infer the number of haplotypes and to match the contigs to their originating haplotypes. Thus, there is a need to cluster the contigs into different groups so that each group represents a haplotype. This step is called contig scaffolding or binning and has been applied for bacterial strain characterization.