A binning tool to reconstruct viral haplotypes from assembled contigs

Jiao Chen; Jiayu Shang; Jianrong Wang; Yanni Sun

Unlike many existing contig clustering tools, our clustering is not applied to a complete contig. Because each contig can contain both haplotype-specific region and shared regions among different haplotypes, using the read coverage profile of the whole contig will confuse the clustering algorithm and makes the convergence slow or leads to wrong assignment of the objects. For example, in Fig. 2 .(c), the contig "A S 2 " contains mutation A from one haplotype and also a shared region S 2 . Thus, significantly more reads will be mapped to the shared region and make the coverage for this contig highly heterogeneous. Thus, the objects as input to the clustering algorithm are "sub-contigs" in windows of depth N , where the sub-contigs are substrings of the contigs in these windows. They are more likely to represent the relative abundance of one haplotype.