Bayesian mixture analysis for metagenomic community profiling.

Sofia Morfopoulou; Vincent Plagnol

The remaining reads are assembled into contigs using the Velvet short read assembler (Zerbino and Birney, 2008) . For each contig we record the number of reads required for its assembly, using this information at the stage of species abundance estimation. A Velvet tuning parameter is the user defined k-mer length that specifies the extent of overlap required to assemble read pairs. Metagenomic assembly is not a straightforward task, as short k-mers work best with the low abundance organisms, while long k-mers with the highly abundance ones. The shorter the k-mer the greater the chance of spurious overlaps, hence we choose relatively high k-mer length, in order to avoid chimeric contigs.