Table 2.
Comparison between methods.
A | |||||||||
Method | Clover | ModuleSearcher(A*) | ModuleDigger | ||||||
Running time | 1.6 min | 0.5 min | 10 s | ||||||
NM | SRR | Sn | NM | SRR | Sn | NM | SRR | Sn | |
OCT4, SOX2, NANOG | 0 | 0 | 0 | 2.4 | 10 | 1.1% | 6 | 10 | 27.6% |
OCT4, SOX2 | 3.9 | 10 | 45.3% | 0 | 0 | 0 | 2 | 10 | 49.1% |
OCT4, NANOG | 0 | 0 | 0 | 0 | 0 | 0 | 3 | 10 | 42.2% |
SOX2, NANOG | 0 | 0 | 0 | 2.4 | 10 | 0.9% | 9 | 10 | 28.4% |
B | |||||||||
Method | Clover | ModuleSearcher (A*) | ModuleDigger | ||||||
Running time | 4 min | 0.5 min | 20 s | ||||||
NM | SRR | Sn | NM | SRR | Sn | NM | SRR | Sn | |
OCT4, SOX2, NANOG | 0 | 0 | 0 | 0 | 0 | 0 | 5 | 10 | 28% |
OCT4, SOX2 | 6.8 | 10 | 45.3% | 21 | 5 | 2.8% | 18 | 10 | 49% |
OCT4, NANOG | 0 | 0 | 0 | 0 | 0 | 0 | 23 | 10 | 42% |
SOX2, NANOG | 0 | 0 | 0 | 21 | 4 | 0.5% | 20 | 10 | 9% |
(A) For all algorithms we used as input the benchmark set, the PWMs of OCT4, SOX2, NANOG and one random PWM. NM: number of modules present in the output for those runs where the RR = 1 (average over runs where RR = 1). SRR (summation of recovery rate): number of runs for which the output contained a module corresponding to the valid modules (OCT4, SOX2, NANOG or combinations thereof). Sn: number of genes containing the valid module in the output (average of runs for which RR was 1). (B) Similar to (A) but using OCT4, SOX2, NANOG in combination with 7 randomly selected PWMs.