File:Strong-Purifying-Selection-at-Synonymous-Sites-in-D.-melanogaster-pgen.1003527.g002.jpg

English: For each 4D site in a conserved amino acid, we use GERP to infer the number of substitutions that have occurred at that site across the Drosophila tree (removing D. melanogaster and D. willistoni from the analysis). We define eight rate classes defined by the number of inferred substitutions across the tree - a proxy for the rate of evolution at the site - and bin the 4D sites accordingly. The class of the slowest evolving sites consists of those codons completely conserved across the ten Drosophila species (0 inferred substitutions along the tree at the 4D site). The fastest evolving class meanwhile has sites with greater than or equal to 9.3 substitutions per site. The remaining substitution classes are spread at intermediate values with a view to roughly equilibrate the number of sites in each class. The substitution bins (b) are as follows: (b1 = 0, 0<b2≤1.4, 1.4<b3≤1.92, 1.92<b4≤3.10, 3.10<b5≤4.40, 4.40<b6≤6.20, 6.20<b7<9.30, b8≥9.30). 10 bootstraps were done for the 4D sites within each bin and their short introns partners. Error bars represent the s.e. of the estimates.