Table 3. Summary of cMonkey2 module construction performance on H. sapiens lung squamous cell carcinoma data set, for several different training schemes, including novel Weeder-based 3’ UTR motif detection, and pre-computed miRNA target gene predictions via novel set-enrichment scoring function. See Methods for details.
| Training Method | Number of Significant Motifs | Comparison p-value(1) | miR2Disease miRNAs | Comparison p-valuea |
|---|---|---|---|---|
| Expression only | 9.3 ± 2.1 | - | 3.5 ± 0.84 | - |
| Weeder motif training | 16 ± 3.1 | 1.5 × 10−3 | 8.8 ± 1.5 | 6.0 × 10−5 |
| Set-enrichment training | 9.5 ± 2.3 | 9.0 × 10−2 | 8.0 ± 0.63 | 1.8 × 10−6 |
Notes: (1) All p-value comparisons are relative to ‘Expression only.’