Formation of Regulatory Modules by Local Sequence Duplication
Figure 3
Common vs. independent descent of binding sites in fly and yeast.
(a) Histogram of the duplication score for 306 pairs of binding sites with a mutual distance of up to 50 bp in the genome of D. melanogaster (sum of grey-shaded and violet-shaded part). Decomposition of counts according to the mixed-descent model (see Materials and Methods):
of the site pairs are of independent descent and have the score distribution
(obtained from pairs with relative distance
bp, dashed line),
of the site pairs of are of common descent and have the score distribution
(violet-shaded). (b) Relative log-likelihood for duplication per site pair, i.e., relative entropy
obtained from the score distribution
of site pairs in the relative distance range
bp (evaluated from a total of 506 sites). The rapid decay of this function suggests a local mechanism generating excess similarity between adjacent sites. (c) Histogram of partial score averages
for all factor pairs
binding the site pairs of (a) (grey-shaded) and corresponding distribution of averages obtained after scrambling the score values of site pairs (normalized to the same number of total counts, dashed line). The two distributions are statistically indistinguishable (KS-test p-value = 0.8378), which shows that positive duplication scores are not limited to a subset of factor pairs. (d) Histogram of the duplication score
for 833 pairs of binding sites with a mutual distance of up to 50 bp in the genome of S. cerevisiae (grey-shaded). The distribution is not significantly different from the null distribution obtained from random site pairs (normalized to the same number of total counts, dashed line), i.e., there is no evidence for common descent as prevalent evolutionary mode.