EVmutation

EVmutation predicts the effects of genetic mutations by modeling residue interdependencies to quantify mutational impacts and capture epistatic interactions across sequences.

Key Features:

• Unsupervised statistical methodology: Employs an unsupervised statistical approach to infer constraints from sequence data without supervised training.
• Residue dependency analysis: Explicitly models interdependencies of residues or bases to capture epistatic interactions between sequence positions.
• Validation against experimental data: Validated by comparison to high-throughput mutagenesis experiments and measurements of human disease mutations, showing improved performance over methods that ignore epistasis.
• Pre-computed human predictions: Provides pre-computed mutation effect predictions for approximately 7,000 human proteins.
• Scalability to large datasets: Designed to handle large sequence datasets derived from high-throughput experimental technologies.

Scientific Applications:

• Quantitative mutation assessment: Quantifies effects of single and combinatorial mutations across genes from any organism.
• Human protein variant interpretation: Supports interpretation of human disease-associated mutations using pre-computed predictions for ~7,000 proteins.
• Epistasis and interaction studies: Enables investigation of epistatic interactions and residue interdependencies relevant to functional and structural analyses.
• Translational research: Informs applied areas such as drug development and personalized medicine by predicting mutation impacts on phenotype.

Methodology:

Uses an unsupervised statistical modeling framework that captures residue dependencies between sequence positions to infer mutational effects and epistasis.