PyRod -- Tracing Water Molecules in Molecular Dynamics Simulations

TL;DR

PyRod is an open-source tool that analyzes water molecules in molecular dynamics simulations to generate pharmacophore models, enhancing structure-based virtual screening without needing ligand structures.

Contribution

PyRod introduces a novel method for deriving pharmacophore features from water molecule behavior in MD simulations, aiding drug discovery efforts.

Findings

Successfully applied to 5 drug targets.

Achieved an early enrichment factor of 54.6 for HIV1 protease.

Provides protein-based pharmacophores without ligand data.

Abstract

Ligands entering a protein binding pocket essentially compete with water molecules for binding to the protein. Hence, the location and thermodynamic properties of water molecules in protein structures have gained increased attention in the drug design community. Including corresponding data into 3D pharmacophore modeling is essential for efficient high throughput virtual screening. Here, we present PyRod, a free and open-source python software that allows for visualization of pharmacophoric binding pocket characteristics, identification of hot spots for ligand binding and subsequent generation of pharmacophore features for virtual screening. The implemented routines analyze the protein environment of water molecules in molecular dynamics (MD) simulations and can differentiate between hydrogen bonded waters as well as waters in a protein environment of hydrophobic, charged or aromatic atom groups. The gathered information is further processed to generate dynamic molecular interaction fields (dMIFs) for visualization and pharmacophoric features for virtual screening. The described software was applied to 5 therapeutically relevant drug targets and generated pharmacophores were evaluated using DUD-E benchmarking sets. The best performing pharmacophore was found for the HIV1 protease with an early enrichment factor of 54.6. PyRod adds a new perspective to structure-based screening campaigns by providing easy-to-interpret dMIFs and purely protein-based 3D pharmacophores that are solely based on tracing water molecules in MD simulations. Since structural information about co-crystallized ligands is not needed, screening campaigns can be followed, for which less or no ligand information is available. PyRod is freely available at https://github.com/schallerdavid/pyrod.