We defined 35,068 coding SNPs (cSNPs), of which 23,089 are silent (synonymous) and 11,979 are missense (non-synonymous) (Supplementary Data 4). Approximately 16% and 4% of non-synonymous SNPs have potentially deleterious effects on protein function as assessed using Polymorphism Phenotyping 2 (PolyPhen 2) (ref. 11) and Sorting Intolerant From Tolerant (SIFT) (ref. 12), respectively (Supplementary Data 4). Approximately 2% (210 SNPs) were defined as deleterious by both algorithms (Fig. 2d and Supplementary Data 4). We identified 58 nonsense SNPs in 53 genes, including 42 stop codon gains and 16 stop codon losses in D2 (Supplementary Data 5). The functional consequence of 210 deleterious missense variants and 58 nonsense variants were further evaluated by comparative genomic analysis (Supplementary Data 6 and 7 and Supplementary Note 1). In addition, 79 missense variants were confirmed by mass spectrometry-based proteomics (Supplementary Data 8 and Supplementary Note 1).
