Multiple property tolerance analysis for the evaluation of missense mutations

Area of Science:

• Bioinformatics
• Computational Biology
• Protein Science

Background:

• Accurate prediction of mutation effects on protein function is vital for clinical diagnostics but current computational methods, particularly sequence alignment-based ones, have limitations.
• Existing methods' accuracy is often dependent on alignment quality and sequence selection, necessitating improvements for reliable clinical application.

Purpose of the Study:

• To develop and evaluate a novel computational method for predicting mutation impact on protein function.
• To enhance prediction accuracy by integrating protein structural information, specifically solvent accessible surface (SAS) properties, into an existing framework.

Main Methods:

• Introduction of Multiple Properties Tolerance Analysis (MuTA), a method based on the conservation of amino acid properties.
• Development of MuTA/S, an extension of MuTA that incorporates the solvent accessible surface (SAS) property.
• An intuitive strategy is employed to group protein residues and adjust property analysis within each group, avoiding complex machine learning or mathematical combinations.

Main Results:

• MuTA demonstrates performance comparable to the established SIFT algorithm for mutation prediction.
• MuTA/S significantly outperforms both SIFT and MuTA, achieving a 2%-25% increase in prediction accuracy across tested proteins (LacI, lysozyme, HIV protease).
• Incorporating the SAS property in MuTA/S substantially reduces the dependency of prediction accuracy on sequence alignments.

Conclusions:

• MuTA/S offers a more accurate and robust approach to predicting mutation effects on protein function compared to existing methods.
• The MuTA/S strategy provides a flexible framework for integrating diverse structural features and biological knowledge, paving the way for more precise mutation impact predictions.
• This method holds promise for improving the reliability of computational predictions in clinical diagnostics and biological research.