Residue Interaction Networks in PfDHFR: Testing an Epistatic "Ridge" Hypothesis

Overview

Fitness-landscape modelling of PfDHFR points to a small set of "ridge" residues — network hubs proposed to act as compensatory or robustness elements that channel evolution along predictable mutational routes. This project tests that hypothesis directly against structural dynamics: building time-resolved residue interaction networks from molecular dynamics trajectories of wild-type and mutant PfDHFR, then asking whether those same positions behave like structural hubs.

Each genotype's MD trajectory is converted into a series of residue–residue interaction graphs (contact-occupancy, dynamic cross-correlation, and hybrid networks), across which graph-theoretic properties — centrality, community structure, embeddings, and robustness to simulated "attack" — are compared across wild type and resistance-associated mutants.

Key Highlights

• Time-resolved residue interaction networks built from all-atom MD across multiple PfDHFR genotypes
• Multiple graph representations: contact-occupancy, dynamic cross-correlation, and hybrid networks
• Centrality, community detection, graph embeddings, and network robustness/attack-tolerance analysis
• Permutation-based null models to statistically test observed network differences
• Directly tests the epistatic "ridge" concept from the fitness-landscape model against real structural dynamics
• Feeds forward into graph-based deep learning approaches for the folate metabolic pathway

Researcher

Led by Danny Muzata in the Chowdhury Lab, Dept. of Biological Sciences, BITS Pilani Hyderabad Campus — part of the wider PfDHFR fitness-landscape research programme.

Google Scholar  •  ORCID