PNBACE: an ensemble algorithm to predict the effects of mutations on protein-nucleic acid binding affinity

Area of Science:

• Computational biology
• Biochemistry
• Bioinformatics

Background:

• Mutations in nucleic acids and proteins can alter protein-nucleic acid binding affinities.
• Previous computational studies have primarily focused on protein mutations, with limited exploration of nucleic acid mutations and generalized prediction methodologies.

Purpose of the Study:

• To develop a generalized computational method for predicting binding affinity changes due to both DNA and protein mutations.
• To assess the impact of nucleic acid mutations on binding affinities.
• To evaluate the applicability of a unified methodology for both mutation types.

Main Methods:

• Development of a generalized algorithm, PNBACE, for predicting binding affinity changes.
• Design of energy-based topological and partition-based energy features.
• Construction of individual prediction models via feature selection.
• Creation of an ensemble model using a differential evolution algorithm.

Main Results:

• PNBACE demonstrates that DNA mutations induce significant binding affinity changes.
• The algorithm can predict the effects of both single-point and multiple-point mutations.
• PNBACE identifies mutations that substantially decrease binding affinities.
• Comparative analyses show PNBACE outperforms existing methods in regression and classification tasks.

Conclusions:

• PNBACE is an effective tool for estimating binding affinity changes in protein-DNA/RNA complexes resulting from mutations.
• The method enhances the understanding of protein-DNA/RNA interactions.
• PNBACE provides a unified approach for analyzing the impact of both nucleic acid and protein mutations on binding affinity.