PyAutoFEP

PyAutoFEP automates Free Energy Perturbation (FEP) calculations to estimate Relative Free Energies of Binding (RFEB) between small molecules and macromolecular targets for drug discovery.

Key Features:

• Automation: Generation of perturbation maps, construction of dual topologies for ligand pairs, system building for molecular dynamics (MD) simulations, and subsequent analysis.
• Force field support: Compatibility with Amber03/GAFF, CHARMM36m/CGenFF, and OPLS-AA/M/LigParGen force fields.
• Enhanced sampling methods: Integration of Replica Exchange with Solute Tempering (REST) and Replica Exchange with Solute Scaling (REST2).
• Flexible λ windows: Support for flexible λ values along perturbation windows to tune alchemical transformations.

Scientific Applications:

• Benchmarking on Farnesoid X receptor ligands: Applied to 14 Farnesoid X receptor ligands in the Drug Design Data Resource Grand Challenge 2, achieving an 88% mean correct sign prediction rate and 75% of predictions with error below 1.5 kcal/mol.
• Validation results: Reported Pearson correlation coefficients of 0.71 ± 0.13 with Amber03/GAFF and 0.66 ± 0.20 with OPLS-AA/M/LigParGen; application of REST2 improved CHARMM36m/CGenFF results to Pearson's r = 0.43 ± 0.21 with minimal impact on other force fields.
• Lead optimization assessment: Estimated probabilities of identifying molecules with pKi one unit better than a lead, with up to sevenfold increases over random sampling when screening 100 molecules.

Methodology:

Generation of perturbation maps, construction of dual ligand topologies, system building for MD simulations, use of Amber03/GAFF, CHARMM36m/CGenFF, and OPLS-AA/M/LigParGen force fields, application of REST and REST2 enhanced sampling, flexible λ windows, and subsequent analysis.