Enriching stabilizing mutations through automated analysis of molecular dynamics simulations using BoostMut

Area of Science:

• Protein Engineering
• Computational Biology
• Biophysics

Background:

• Thermostability is crucial for protein applications, but identifying stabilizing mutations is challenging.
• Current methods like molecular dynamics (MD) simulations and visual inspection for mutation filtering are slow and subjective.
• Automated prediction of stabilizing mutations is needed to improve protein engineering workflows.

Purpose of the Study:

• To introduce Biophysical Overview of Optimal Stabilizing Mutations (BoostMut), a computational tool for standardized and automated mutation filtering.
• To formalize and automate the analysis of dynamic structural features from MD simulations for protein stability assessment.
• To improve the success rate of identifying stabilizing mutations in protein engineering.

Main Methods:

• Development of BoostMut, a computational tool analyzing dynamic structural features from MD simulations.
• Standardization and automation of mutation filtering principles previously used in manual verification.
• Benchmarking BoostMut across multiple datasets and integrating it with a machine learning model for enhanced performance.

Main Results:

• BoostMut provides consistent and reproducible protein stability assessments.
• Integrating BoostMut's biophysical analysis improved prediction rates across various thermostability predictors.
• Experimental validation on limonene epoxide hydrolase identified novel stabilizing mutations, increasing the overall success rate.

Conclusions:

• BoostMut offers a standardized, automated, and effective approach to filtering stabilizing mutations.
• The tool enhances existing thermostability prediction workflows and can aid in data labeling for future predictor training.
• BoostMut facilitates the discovery of overlooked stabilizing mutations, advancing protein engineering for biocatalysts and biomedicines.