EASE-MM: Sequence-Based Prediction of Mutation-Induced Stability Changes with Feature-Based Multiple Models.

Area of Science:

• Computational biology
• Protein engineering
• Bioinformatics

Background:

• Accurate prediction of protein stability changes (ΔΔGu) is crucial for protein engineering and characterizing non-synonymous single nucleotide variants (SNVs).
• Existing methods often lack accuracy or require complex structural information.

Purpose of the Study:

• To develop and validate a novel, sequence-based prediction method for protein stability changes (ΔΔGu) induced by single amino acid substitutions.
• To assess the performance of the new method against existing sequence-based and structure-based approaches.

Main Methods:

• Developed EASE-MM, a consensus method using five specialized support vector machine (SVM) models.
• Selected final prediction based on secondary structure and accessible surface area of mutated residues.
• Applied EASE-MM to single-domain monomeric proteins using only sequence and mutation data.

Main Results:

• EASE-MM achieved a Pearson correlation coefficient of 0.53-0.59 in cross-validation and independent testing.
• Outperformed other sequence-based prediction methods and showed comparable or better performance than structure-based energy functions.
• Demonstrated that disease-causing variants are associated with larger predicted ΔΔGu magnitudes using EASE-MM on human germline SNVs.

Conclusions:

• EASE-MM provides an accurate and efficient method for predicting protein stability changes from single amino acid substitutions.
• The method has potential applications in understanding the impact of genetic variants on protein function and disease.
• EASE-MM is available as a web server for broader scientific use.