Table 5.
Performance of aligners in mapping simulation reads generated from the unfiltered human genome (repetitive regions were kept)
| Aligner | Art |
Mason |
Our simulator |
Time (min) | |||
|---|---|---|---|---|---|---|---|
| Rec (%) | Acc (%) | Rec (%) | Acc (%) | Rec (%) | Acc (%) | ||
| Subread | 81.5 | 97.9 | 88.8 | 96.1 | 88.5 | 98.0 | 17 |
| Subread -f 100 | 84.4 | 97.7 | 91.5 | 96.0 | 91.3 | 97.9 | 19 |
| Subread -f 200 | 85.5 | 97.6 | 92.5 | 95.9 | 92.4 | 97.8 | 21 |
| Subread -f 300 | 86.1 | 97.5 | 93.1 | 95.8 | 92.9 | 97.7 | 22 |
| Bowtie2 | 87.6 | 95.2 | 95.2 | 95.3 | 95.7 | 96.0 | 83 |
| BWA | 87.1 | 97.2 | 95.5 | 95.7 | 78.6 | 96.4 | 497 |
| Novoalign | 89.8 | 97.3 | 93.5 | 97.1 | 140 | ||
One hundred thousand 100 bp reads were each generated from Art and Mason simulators, and 10 million 101 bp reads were generated from our simulator. ‘-f’ option of Subread specifies the threshold for removing uninformative subread. For example, ‘-f 100’ means those subreads that occur 100 or more times in the reference genome were removed. Running time was measured using our simulation data.