Epistatic Network Constraints Shape Predictable Antifolate Resistance in Plasmodium falciparum Dihydrofolate Reductase

Overview

Antifolate drugs such as pyrimethamine target dihydrofolate reductase (DHFR) in Plasmodium falciparum, the parasite responsible for the deadliest form of malaria. Resistance typically accumulates through a small, well-known set of active-site mutations — but why evolution converges so reliably on the same mutational routes, out of the enormous number of theoretically possible ones, has remained an open question.

This project builds computational models of the PfDHFR fitness landscape across the combinatorial space of resistance-associated mutations. By quantifying epistasis — how the fitness effect of one mutation depends on which others are already present — we show that the accessible mutational network is far more constrained than the full combinatorial space, funnelling evolving populations along a narrow set of high-probability, largely predictable pathways toward resistance.

It is the flagship thread of a broader, multi-pronged PfDHFR research programme: sequence conservation and co-evolution analysis across thousands of DHFR homologues, structural dynamics from molecular dynamics simulations of wild-type and mutant enzymes, and a population-genetic (Wright–Fisher) adaptive-walk simulator that stitches these signals into a composite fitness model. Computational predictions are now being tested against ongoing wet-lab thermal-stability experiments with collaborators at BITS Pilani.

Key Highlights

• Computational fitness-landscape modelling across the PfDHFR resistance-mutation space
• Quantifies epistatic interactions that constrain which mutational orders are viable
• Identifies a narrow set of dominant, predictable adaptive trajectories to resistance
• Integrates sequence co-evolution, structural dynamics, and population-genetic simulation into one fitness model
• Predictions under ongoing experimental validation via thermal-stability assays
• Connects fitness-landscape theory to real-world antifolate drug-resistance surveillance

Related Threads

This research question is approached from several complementary computational angles, each developed as its own project:

• Residue Interaction Networks in PfDHFR — graph-theoretic analysis of MD-derived residue networks testing the epistatic "ridge" hypothesis
• Protein–Ligand Interaction Modelling — graph-based deep learning for antifolate binding prediction
• Structure-Based Virtual Screening of PfDHFR — applying these insights to candidate inhibitor discovery

Lead Author

This is my flagship doctoral research project, led under Danny Muzata, a biomedical data scientist at BITS Pilani, Hyderabad Campus, working on protein evolution and antimicrobial resistance in the Chowdhury Lab.

Google Scholar  •  ORCID