For each translated DamID binding interval dataset, the total number of binding intervals overlapping an intron by at least 100 bp, as well as the number of binding intervals overlapping the first intron of a transcript, were counted.
We also found an association between Dichaete binding and transcription start sites (TSS), observing a peak of binding intervals within 500 bp of a TSS, indicating a preference for this region.
This mode of regulatory evolution appears to be important for Dichaete, as we found that in all pairwise comparisons, the majority of binding intervals that were not positionally conserved in one species had at least one binding interval present at a different position within the same locus in the other species.
We have recently used combinations of genome-wide ChIP and DamID data to generate high confidence in vivo binding profiles and define a set of core binding intervals for both Dichaete and SoxN in Drosophila melanogaster.
However, for both TFs, DamID binding intervals that overlap a core interval are significantly more likely to be conserved than those that do not (Dichaete: χ = 1408.6, d.f. = 3, p = 4.10e-305; SoxN: χ = 733.1, d.f. = 1, p-value = 1.90e-161) (Additional file 8: Figure S4).
After normalisation and correction for multiple hypothesis testing, we identified between 17,000 – 26,000 binding intervals (p < 0.05) for Dichaete in D. melanogaster, D. simulans and D. yakuba, and for SoxN in D. melanogaster and D. simulans.
We employed DiffBind [ 87] to perform a differential enrichment analysis, comparing Dichaete-Dam binding profiles between two species for each binding interval, identifying binding intervals that showed significant quantitative differences between each pair of species [ 87].
These binding intervals were used for further analysis.
Although it is not known whether these differences in Sox motifs found in Dichaete and SoxN binding intervals are responsible for different functions of the two TFs, it is striking that the same patterns appear independently in the genomes of multiple species.
FIMO was used to search for matches to each PWM in all binding intervals, and the resulting hits were used to calculate the average number of motifs per binding interval [ 90].
