Figure 2

Performance of three short-read assemblers on a single-end metatranscriptomic dataset. Three different single-end assemblers (with varying k-mer parameters where appropriate) were applied to the NOD503CecMN single-end dataset and evaluated on the basis of: 1) the probability of contigs of different lengths having significant sequence similarity (bit score >50) to a known protein, as well as the percentage of reads which could be annotated (top panel), and 2) contig length distributions (bottom panel). While the assemblers varied greatly in the contig length distribution, number of contigs assembled, and number of reads which could be matched to an annotated contig, all contigs over 180 bp, irrespective of the assembler used to generate them, had a consistently high probability of having significant sequence similarity to a known protein.