Figure 3

Transcription Factor Binding Site Alignment Accuracy. A: Binding site alignment accuracy varies across tools and divergences. Fraction of binding sites overlapping in four species alignments was measured as a function of total divergence distance. B: Binding sites are often still overlapping in alignments even when they are not perfectly aligned. Fraction of binding sites perfectly aligned in four species alignments subtracted from the fraction of binding sites overlapping in four species alignments was measured as a function of total divergence distance. C: Binding site alignment accuracy is highly correlated with overall alignment accuracy and is consistently higher. Fraction of binding sites overlapping in four species alignments was measured as a function of overall alignment accuracy. D: Binding site alignment accuracy varies across branches in a tree and is best for leaf-to-leaf alignments and worst for node-to-node alignments. Fraction of binding sites overlapping along branches in three and four species trees subtracted from the fraction of binding sites overlapping in two species Clustalw alignments, where the divergence along each branch is the same, was measured as a function of divergence distance. E: Binding site alignment accuracy is positively correlated with binding site density in an enhancer. Fraction of binding sites overlapping in replicate four species Mlagan alignments of each of the 36 enhancers was measured as a function of the density of binding sites in the enhancer. F: Binding site alignment accuracy is positively correlated with binding site length. Fraction of binding sites overlapping in four species Mlagan alignments for each of the eight transcription factors was measured as a function of the length of the transcription factors' binding sites.