MetaCarvel: linking assembly graph motifs to biological variants

Reconstructing genomic segments from metagenomicsdata is a highly complex task. In addition to general challenges, such as repeats and sequencing errors, metagenomicassembly needs to tolerate the uneven depth of coverage among organisms in a community and differences between nearly identical strains. Previous methods have addressed these issues by smoothing genomic variants. We present a variant-aware metagenomicscaffolder called MetaCarvel, which combines new strategies for repeat detection with graph analytics for the discovery of variants. We show that MetaCarvel can accurately reconstruct genomic segments from complex microbial mixtures and correctly identify and characterize several classes of common genomic variants.