BiMeta

A two-phase algorithm for the binning of metagenomic reads without using reference genomes. Instead of directly clustering reads, the main idea of BiMeta is to provide an additional preprocessing phase in which reads potentially belonging to the same cluster are grouped and each group is presented by a so-called seed of non-overlapping reads. The idea is motivated by a careful observation of the l-mer frequency distributions on sets of non-overlapping reads extracted from microbial genomes. BiMeta demonstrates to be able to achieve higher performance than the state-of-the-art binning algorithms on both simulated and real metagenomic datasets. Another strength of BiMeta is that it can work well with both short and long reads.