PDA-PreD

PDA-PreD predicts binding affinities of protein-DNA complexes from structural features to quantify determinants of binding free energy (ΔG).


Key Features:

  • Data Collection: Uses a dataset of experimental binding free energy (ΔG) values for 391 protein-DNA complexes.
  • Structural Feature Analysis: Derives structure-based features including interaction energy, contact potentials, volume and surface area of binding site residues, DNA base-step parameters, and atom-type contacts between proteins and DNA.
  • Correlation with Binding Affinity: Identifies factors influencing binding affinity such as number of DNA strands, functional and structural classes of proteins, and binding-site properties (atom contacts, binding site volume, interaction energies, contact potentials).
  • Predictive Modeling: Employs multiple regression equations tailored to different structural and functional classes of protein-DNA complexes.
  • Performance Metrics: Reports an average correlation coefficient of 0.78 and a mean absolute error of 0.98 kcal/mol between experimental and predicted affinities as validated by jack-knife testing.

Scientific Applications:

  • Recognition Mechanism Analysis: Elucidates structural determinants of protein-DNA recognition by relating structural features to binding free energy.
  • Experimental Design and Intervention: Informs experimental design and interventions targeting gene regulation by predicting effects on binding affinity.
  • Drug Discovery: Supports drug discovery efforts aimed at modulating protein-DNA interactions by providing affinity predictions for complexes of interest.

Methodology:

Collects experimental binding free energies (ΔG) for 391 protein-DNA complexes, extracts structure-based features (interaction energy, contact potentials, binding-site volume and surface area, DNA base-step parameters, atom-type contacts), analyzes feature–affinity relationships, trains regression models using machine learning techniques with multiple regression equations, and validates performance via jack-knife testing.

Topics

Details

Cost:
Free of charge
Tool Type:
web application
Operating Systems:
Mac, Linux, Windows
Programming Languages:
Python
Added:
11/7/2023
Last Updated:
11/24/2024

Operations

Publications

Harini K, Kihara D, Michael Gromiha M. PDA-Pred: Predicting the binding affinity of protein-DNA complexes using machine learning techniques and structural features. Methods. 2023;213:10-17. doi:10.1016/j.ymeth.2023.03.002. PMID:36924867. PMCID:PMC10563387.

PMID: 36924867
Funding: - National Institutes of Health: R01GM123055, R01GM133840 - Ministry of Science and Technology, Government of Nepal: OVDF 2022 - Science and Engineering Research Board: CRG/2020/000314 - National Science Foundation: CMMI1825941, DBI2003635, DBI2146026, DMS2151678, IIS2211598, MCB1925643

Links