Disease-causing mutations have highest scores SNAP predicted the impact of function for five different data sets of point mutations: disease related + observed effect and disease related mutants, mutations with observed effect, unknown disease relation, and random mutations. For each set we display the predicted functional severity of mutations. (A) Scores above zero (horizontal line) correspond to effect, scores below to neutral, the distance from 0 correlates to severity; lower/upper bound and bar in the box represent the lower/upper quartile and median. 90% of disease related+observed effect and over 86% of the disease related mutations were predicted to effect function, compared to only 51% in mutations of unknown disease relation. Effect predictions dominated the observed effect mutants less (76%) than the disease related mutants (86%). The effect in random mutations (44%) provided an upper bound for effect mutations in proven non-disease related variants. (B) Cumulative distributions of predicted functional severity; points on a curve correspond to fractions (y-axis) of mutations with SNAP scores (x-axis) ≥ this value. The vertical line separates neutral from effect. Disease-causing mutations were predicted to be most severe (black solid and dashed lines above all others). These results suggest that change in function may explain most disease-related mutations.
Schaefer et al. BMC Genomics 2012 13(Suppl 4):S11 doi:10.1186/1471-2164-13-S4-S11