Virtual Screening - Usage Guide

Overview

Screen compound libraries against protein targets using molecular docking with AutoDock Vina. Predict binding poses and affinities for hit identification.

Prerequisites

pip install vina
pip install rdkit meeko

Quick Start

Tell your AI agent what you want to do: - "Dock my compound library against this protein" - "Set up a virtual screen against the binding site" - "Rank compounds by predicted binding affinity" - "Prepare my protein for docking"

Example Prompts

Receptor Preparation

"Prepare my protein PDB for docking by reviewing waters, cofactors, metals, and alternate locations; retain justified structural components and add hydrogens."

"Convert my protein to PDBQT format for Vina."

Docking

"Dock this ligand to my protein and show the best poses."

"Screen my library of 1000 compounds against the active site."

Analysis

"Rank my docked compounds by binding affinity."

"Extract the top 10 hits with affinities better than -8 kcal/mol."

What the Agent Will Do

1. Prepare receptor (review waters, cofactors, metals, and alternate locations; retain justified structural components; add H and convert to PDBQT)
2. Prepare ligands (generate 3D, minimize, convert to PDBQT)
3. Define binding site from ligand or coordinates
4. Run Vina docking for each compound
5. Collect and rank results by affinity

Tips

• AutoDock Vina 1.2.x is an open baseline; benchmark any GPU port on the same hardware, target, library tranche, and search settings before adopting it
• Set the box from the known ligand/pocket and intended ligand-size range; confirm that ligands can translate and rotate inside it instead of enforcing a universal 30 A limit
• Vina's documented default exhaustiveness is 8. Use it as a baseline, then increase effort geometrically and compare redocking recovery, score/pose stability, and runtime on target-relevant controls; values such as 32 or 64 are test points, not universal production settings
• Center binding box on co-crystallized ligand if available
• Review waters, cofactors, metals, alternate locations, missing atoms, and protonation explicitly; retain conserved structural waters when they mediate binding
• Note: Vina 1.1.2 vs 1.2 may give different poses

Related Skills

• molecular-io - Load and convert molecules
• admet-prediction - Filter before docking
• structural-biology/structure-io - Protein structure handling
• structural-biology/modern-structure-prediction - Generate targets