Decoding Missense Variants by Incorporating Phase Separation via Machine Learning

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology
• Genetics

Background:

• Predicting the functional impact of protein variants is crucial but challenging, especially for variants of uncertain significance (VUS) in intrinsically disordered regions (IDRs).
• Intrinsically disordered regions are implicated in numerous physiological processes, and their dysfunction is linked to various diseases.
• Phase separation, a process central to many cellular functions, is closely associated with IDRs.

Purpose of the Study:

• To develop a computational method for predicting the impact of missense mutations on protein phase separation.
• To improve the interpretation of variants of uncertain significance (VUS) located in intrinsically disordered regions (IDRs).
• To aid in understanding the pathogenicity of disease-associated variants.

Main Methods:

• Developed PSMutPred, a machine learning model leveraging missense variants that alter phase separation propensity.
• Integrated phase separation principles into the analysis of missense variants within IDRs.
• Validated PSMutPred predictions using in vitro experiments.

Main Results:

• PSMutPred demonstrates robust performance in predicting missense variants that affect natural phase separation.
• The model successfully predicted the impact of mutations on phase separation propensity.
• In vitro experiments confirmed the predictive accuracy of PSMutPred.

Conclusions:

• PSMutPred significantly contributes to decoding the pathogenesis of disease variants, particularly those in IDRs.
• The approach aids in the interpretation of a large number of VUSs in IDRs, accelerating clinical diagnosis.
• This work highlights the importance of phase separation in understanding variant effects and disease mechanisms.