Disentangling the contribution of each descriptive characteristic of every single mutation to its functional effects

Mutational effects predictions continue to improve in accuracy as advanced artificial intelligence (AI) algorithms are trained on exhaustive experimental data. The next natural questions to ask are if it is now possible to gain insights into which attribute of the mutation contributes how much to the mutational effects, and if one can develop universal rules for mapping the descriptors to mutational effects. In this work, we mainly address the former aspect using a framework of interpretable AI. Relations between the physico-chemical descriptors and their contributions to the mutational effects are extracted by analyzing the data on 29,832 variants from 8 systematic deep-mutational scan studies. It is found that the intuitive dependences of fitness and solubility on the distance of the amino acid from active site could be extracted and quantified. The dependence of the mutational effect contributions on the number of contacts an amino acid has or the BLOSUM score descriptor of the change showed universal trends. Our attempts in the present work to explain the quantitative differences in the dependence on conservation and SASA across proteins were not successful. The work nevertheless brings transparency into the predictions, development of rules, and will hopefully lead to uncovering the universalities among these rules.