PEMPNI

PEMPNI predicts the impact of missense mutations on protein-DNA and protein-RNA interactions to quantify changes in binding affinity for MPDs and MPRs and inform disease-related effects.

Key Features:

• Dual-module approach: Integrates energy-based and nonenergy-based feature modules to improve prediction accuracy.
• Geometric partition-based energy features: Uses geometric partitioning to derive energy features that capture changes in binding free energy caused by mutations.
• Interface-based structural features: Extracts structural and sequence-derived features from protein–nucleic acid interfaces.
• Separate models for MPDs and MPRs: Develops distinct computational frameworks tailored to mutations affecting DNA-binding proteins (MPDs) and RNA-binding proteins (MPRs).
• Regression algorithms: Employs separate regression algorithms for each mutation class to reflect their unique tendencies in altering binding affinities.
• Ensemble learning: Incorporates ensemble learning techniques within the regression frameworks.
• Rigorous feature selection: Applies feature selection procedures to identify predictive features for each module and class.

Scientific Applications:

• Predicting mutation effects on protein-DNA interactions: Estimates how missense mutations alter binding affinity between proteins and DNA (MPDs).
• Predicting mutation effects on protein-RNA interactions: Estimates how missense mutations alter binding affinity between proteins and RNA (MPRs).
• Assessing binding free energy changes: Quantifies mutation-induced changes in binding free energy at protein–nucleic acid interfaces.
• Interpreting disease mechanisms: Supports investigation of how mutations in DNA- and RNA-binding proteins contribute to disease-related functional changes.

Methodology:

Combines a geometric partition-based energy module and an interface-based structural (nonenergy) module, and applies separate regression algorithms with ensemble learning and rigorous feature selection to build distinct models for MPDs and MPRs.