Fig. 2

Xander gene assembly workflow. Two types of input sequences are required: one or more metagenomic read files used to build the de Bruijn graph and one set of reference sequences for each targeted gene, for building specialized profile HMMs using a modified version of HMMER 3.0 (see the “Implementation” section). During the search phase, Xander uses a combined weighted assembly graph to assemble genes (contigs). After assembly, several filters are applied at the quality filter step: chimeric genes, or genes below length cutoff or HMM score cutoff are discarded, and genes are clustered at 99 % aa identity and the longest one from each cluster is chosen as the representative. The quality-filtered genes are further processed to provide coverage and abundance information