Table 2. Enriched TFBSs in core HIF binding regions (variable selection).
Overrepresented PWM | Number chosen | Transcription factor |
MA0259.1 | 100 | HIF1 |
PF0146 | 86 | unknown (RRCCGTTA) |
PF0032 | 82 | NFE-2 |
M00222 | 80 | Hand1:E47 |
M00188 | 77 | AP1 |
PF0096 | 67 | unknown (YGCANTGCR) |
PF0014 | 58 | ATF3 |
M00109 | 57 | CEBPB |
M00185 | 45 | NFY |
M00246 | 36 | EGR2 |
M00244 | 20 | NGFIC |
MA0154.1 | 15 | EBF1 |
PF0009 | 13 | ATF3 |
M00302 | 11 | NFAT |
M00002 | 10 | E47 |
The Table indicates sequence motifs/PWMs identified as discriminative of core HBRs employing correlation-based feature selection in 10 iterations of 10-fold stratified cross-validation. The results are ranked according to the total number of folds (up to a hundred) in which the variable was chosen by the algorithm (Number chosen). The associated transcription factor, were known, is indicated along with the PWM (Tr. factor). Predictions coincident with Fisher’s exact test (Table 1) are highlighted in bold.