Fig. 3: Robustness of aligners to sequence divergence.
From: Efficient sequence alignment against millions of prokaryotic genomes with LexicMap
This is measured by simulating 250-bp, 500-bp, 1,000-bp and 2,000-bp reads with coverage of 30× from ten bacterial genomes, adding mutations to achieve sequence divergence between 0% and 20% and then aligning back to the source genome. COBS is a k-mer index; thus, we show what proportion of the reads were detected as being present in the source genome (rather than being aligned). This falls off very rapidly as similarity drops because each base disagreement loses an entire 31-mer. For the other tools, we measure the proportion of reads that are correctly aligned back to the source genome. BLASTn (ws = 15) represents BLASTn with a word size of 15. BLASTn (with no brackets) refers to the default setting of BLASTn, with word size of 28. All data are available in Supplementary Table 6.
