Figure 1.
COPS first scans the input sequences using all known TF binding motifs annotated in open source databases or retrieved from other resources and builds the frequent pattern (FP)-tree. The statistical significance (Z score) of the motif co-occurrences is calculated by comparing the log likelihood score of the frequent pattern to the log likelihood score distribution of the background. The percentage of overlap between the motifs of the pair is also calculated and reported. Additionally, COPS offers the option to calculate the preferred distance between co-occurring TF binding motifs and its statistical significance (Z score).
Table 1.
Classification of TFs co-occurring with Twi on Twi-bound genomic regions in stage 10 to 11 Drosophila embryos.
Figure 2.
Selected examples of the statistical significance of the identified co-occurrence patterns.
The statistical significance of selected motif pairs from the different genome-wide datasets is depicted. The log likelihood score (blue bar) of the motif pairs Bap/Twi (A), Tin/Twi (B), Ase/Pros (C) and Snail/Pros (D) is shown in relation to the log likelihood score distribution of the background (red curve).
Figure 3.
Observed overlap of the genomic regions bound by the TF Twi and its co-regulators.
A-C: The observed overlap (blue bar) between genomic regions bound by Twi and genomic regions bound by its known co-regulator Snail and its candidate co-regulator Ase is depicted in comparison to the expected random overlap (red curve). The overlap of genomic regions bound by Twi in stage 10 to 11 [6] with the genomic regions bound by Twi/Snail/Dorsal in stage 5–7 [28] (A), Snail (stage 10 to 11 [7]) (B) and Ase (stage 10 to 11 [7]) (C) is shown. A’-C’: Distribution and overlap (red) of Twi- and Twi/Snail/Dorsal- (A’), Twi- and Snail- (B’) and Twi- and Ase- (C’), bound genomic regions. The regions bound by Twi are shown in blue, the regions bound by the co-regulatory TFs in green and the overlapping regions bound by both Twi and the co-regulatory TFs in red.
Table 2.
Overlap of genomic regions bound by Twi and Twi co-regulatory TFs in vivo.
Table 3.
Classification of TFs co-occurring with Pros in DamID identified genomic regions in stage 10 to 11 Drosophila embryos.
Table 4.
Classification of TFs co-occurring with Nkx2.5 in ChIP-on-Chip identified genomic regions in HL-1 cells.
Table 5.
Short distance arrangements between BS of known co-regulatory TFs.
Figure 4.
Comparison of COPS to two other computational approaches.
The plots depict the Matthews Correlation Coefficient (MCC) values as determined using COPS and the published tools CPModule and ModuleDigger, for the co-occurrence patterns Bap/Twi, Tin/Twi, Ase/Pros and Snail/Pros in sequence-sets of increasing size (ranging from 50 to 300 sequences). The MCC values used for evaluating the performance of the different computational tools were calculated as described in the Materials and Methods. The red line illustrates the performance of COPS, the blue line CPModule and the green line ModuleDigger.