Table 2. Comparison of FACS algorithms with CoMeta.
| k | MC | Sensitivity | Precision | Classified | t |
|---|---|---|---|---|---|
| [%] | [%] | [%] | [%] | [hh:mm:ss] | |
| FACS-P | |||||
| 18 | 80 | 97.62 | 97.86 | 99.76 | 00:03:14 |
| 21 | 65 | 97.86 | 98.08 | 99.78 | 00:02:49 |
| 21 | 70 | 97.82 | 98.27 | 99.55 | 00:02:49 |
| 24 | 55 | 97.77 | 98.12 | 99.64 | 00:02:36 |
| 27 | 45 | 97.65 | 98.07 | 99.58 | 00:02:27 |
| FACS-C | |||||
| 17 | 30 | 99.92 | 90.20 | 99.93 | 00:01:08 |
| 17 | 40 | 98.78 | 93.25 | 98.78 | 00:01:12 |
| 19 | 30 | 99.48 | 92.65 | 99.48 | 00:00:49 |
| 21 | 30 | 98.26 | 94.27 | 98.27 | 00:00:43 |
| pre-CoMeta | |||||
| 15 | 55 | 99.30 | 93.56 | 99.31 | 00:01:52 |
| 18 | 45 | 99.42 | 93.36 | 99.43 | 00:01:21 |
| 21 | 45 | 99.05 | 93.93 | 99.06 | 00:01:08 |
| 25 | 30 | 99.56 | 92.05 | 99.57 | 00:01:09 |
| 27 | 35 | 99.36 | 93.07 | 99.37 | 00:01:16 |
| CoMeta | |||||
| 18 | – | 97.91 | 97.91 | 100.00 | 00:01:37 |
| 21 | – | 98.40 | 98.41 | 99.99 | 00:01:36 |
| 24 | – | 98.69 | 98.75 | 99.93 | 00:01:37 |
| 27 | – | 98.71 | 99.08 | 99.63 | 00:01:30 |
Comparison of the best classification results obtained using four methods (bold values indicate the best score for each column):
FACS-P: the FACS 2.1 program in Perl [49]. When read is classified to some G i-th reference sequence, it does not be compared with any further reference sequence;
FACS-C: the FACS program in C, which was downloaded from https://github.com/SciLifeLab/facs. The reads are classified to each reference sequence to which similarity is highest than MC;
pre-CoMeta: the only comparison step of CoMeta algorithm (without assignment). This is a similar strategy as implemented in FACS-C.
CoMeta: the full proposed algorithm, the reads are classified to the reference sequence according to the highest score.