Table 2.

Overview of computational studies on the evolutionary impact of exonic splice regulatory information

References	Motif density	d S/d 4 decrease in motifs	Over-all d S/d 4 decrease	Motifs	Control
Parmley et al. (2006)	~30.42%	~8.19% (including CpG sites)/11.03% (excluding CpG sites) (alignment to mouse)	~2.49% (including CpG sites)/3.36% (excluding CpG sites)	238 RESCUE-ESE ESE hexamers (Fairbrother et al. 2002)	Non-ESE sites
Cáceres and Hurst (2013)	13.1–32.7% (exon ends only)	8.5–17.1% (exon ends only, alignment to mouse)	1.2–4% (extrapolated from exon ends to the full sequence)	Various sets of putative ESEs, formed by taking intersections of pre-existing sets	Either all non-ESE sites near exon ends or sites overlapping with nucleotide-matched control motifs
Savisaar and Hurst (2017)	~57.3%	~4.1% (alignment to macaque)	~2.4%	1483 motifs experimentally determined to be recognized by human RBPs	Sites that overlap dinucleotide-matched control motifs

For Cáceres and Hurst (2013), the figures are presented as a range, as they depend on the set of motifs and the method of control used. Note that some studies considered d _S (rate of evolution at synonymous sites) while others considered d ₄ (rate of evolution at fourfold degenerate sites). Parmley et al. (2006) also provided a second estimate for the over-all decrease in d _S (~8%), however, only the lower estimate is reproduced here because of concerns that the reasoning used to derive the higher value may have been circular

ESE exonic splice enhancer, RBP RNA-binding protein