Using RINs to understand cancer mutations: deleterious mutations are more commonly associated to highly connected amino acids
Residue Interaction Networks. Mutation effects. Driver and passenger mutations. Mutation predictors.
Motivation: Many efforts to identify cancer driver mutations have been made, most at the sequence level. The understanding of the structural impact of a given amino acid change is of major importance to cancer medical research. Here, we used residue interaction networks (RINs) parameters to a large-scale analysis of missense mutations from 16 cancer types, allowing us to infer their respective structural effects and to verify if changes in highly connected amino acids are more likely to give rise to driver mutations.
Results: We used RINs to analyze which network parameters are more common in sequence sites (node) with occurrence of reported missense cancer mutations. The distribution of mutations’ count per node degree varies significantly from random simulations and also from the distribution of a dataset of human single nucleotide polymorphisms (SNPs). Also, the proportion of deleterious mutations was significantly increased in nodes with high degree of connectivity, when used two different criteria for its classification: proportions of software predictors (NDamage) and ClinVar database classification. Therefore, taking into account these results, we can conclude that changes in highly connected amino acids are indeed more likely to give rise to drivers mutations, since its higher proportion of occurrence in these nodes, and that RINs analysis can be used as an additional parameter to aid the prediction of cancer drivers mutations.
Availability: The R scripts, Python scripts and SQL instructions used here are available in (...) GitHub page.