A binning tool to reconstruct viral haplotypes from assembled contigs

Jiao Chen; Jiayu Shang; Jianrong Wang; Yanni Sun

For the shortest contig set, MaxBin only reported two clusters with one contig for each cluster, leaving 59 (96.7%) contigs unclassified. For contigs sets from 2000 -5000, MaxBin was able to generate five clusters, but with~30% contigs unclassified. The results of MaxBin usually have lower precision or recall values than VirBin. In addition, for contig sets from 1000 -4000, there are haplotypes without correctly assigned contigs. The lower sensitivity of MaxBin could be caused by its dependency on both sequence composition and contig coverage for clustering. Due to high sequence similarities between viral haplotypes, sequence composition is not informative enough in differentiating contigs from different viral strains. Instead, MaxBin could mistakenly cluster contigs from the homogeneous regions of the viral genome, leading to more chimeric clusters.