Table 3. Significant motifs and motif pairs for analysis with known and predicted matrices
Pair IDSignificant motifs and motif pairsP‐valueFit coefficientsTarget gene countTissue with maximum expressionSpecificity in liver
ModelGenome‐wide
No. of target genesPercentage countNo. of target genesPercentage count
HNF‐13.0 × 10−15182.8101.0940.6Liver2.4 × 10−15
P1TGACCTTTG*HNF‐10.0023521.9101.01541.0Liver3.8 × 10−13
P2TGACCTTTG*ACCCTAGACC0.0002−348.2545.48125.3Monocytes
P3ARP‐1*ATGGAAAGA0.0006299.2171.71541.0Liver6.7 × 10−6
P4TTAACATGCA(∼FXR IR‐1)*ATGGAAAGA0.0004−70.222222.2341622.3B cells
P5AATTGAAT(∼LEF‐1)*ATGGAAAGA3.5 × 10−7−66.033133.1493932.3Monocytes
P6ATGGAAAGA*ACCCTAGACC0.000525.849949.9724447.3Monocytes
P7TGACCTTTG*ATGCCTGTC(∼USF‐2)6.1 × 10−1021.056156.1848155.4Monocytes
P8TGAATGAAT*HNF‐13.0 × 10−15−18.698998.915 21299.4B cells
  • Predicted matrices are represented by their consensus sequences. Matches to known PWMs are indicated in parentheses. For example, the PWM ATGCCTGTC matches USF‐2. Similarity was determined using MatCompare (Schones et al, 2005). USF‐2 is a ubiquitous factor, FXR is a bile acid receptor and LEF‐1 is a nuclear mediator in Wnt signaling pathway. ARP‐1 is involved in multiple processes including lipid metabolism. Under target gene count, model refers to the genes used to fit the spline model to microarray data (total=1000), whereas genome‐wide indicates all the genes on the array whose promoters were available (total=15 309). Final spline model and the significant predicted matrices for this analysis are reported in Supplementary note.