liveBook · Manning

Structure-based virtual screening

As explained in Build AI Drug Discovery Pipelines

Structure-based virtual screening prioritizes compounds using structural information about a biological target or binding site.

Structure-based virtual screening (SBVS) prioritizes large compound libraries against a biological target’s three-dimensional structure. It supports early drug discovery by identifying compounds and binding poses for more expensive evaluation. section 9.1

A practical SBVS workflow typically proceeds as follows:

  • Prepare an experimentally determined receptor structure, considering uncertainties such as alternative side chains, missing loops, waters, ligands, and crystallographic artifacts. section 9.1.1
  • Represent the receptor and ligand with explicit three-dimensional atomic coordinates, define a binding box, and score candidate poses on interaction-energy grids. section 9.1.1
  • Rank compounds with a docking score, then inspect pose quality, rescore promising candidates, use short molecular dynamics when appropriate, and confirm results experimentally. section 9.1.1

Docking a compound library

Given a prepared protein structure, a three-dimensional ligand conformer, and a defined binding site, docking produces candidate poses and Vina scores. Lower scores rank more favorably, but they estimate scoring-function preference rather than measured binding free energy or calibrated binding probability. A strong-ranked pose should therefore be compared with a crystallographic pose or consensus rescoring before assay testing. section 9.1.1

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Explains docking inputs, representations, scoring, validation, and failure modes, but stops short of measuring binding free energy directly.
Examples
Uses PDB-derived protein structures, kinase targets, PDBQT coordinates, binding boxes, ligand poses, Vina scores, and crystallographic pose comparisons.
What it assumes
Requires a prepared three-dimensional receptor, ligand conformer, and binding site while accounting for structural uncertainty and receptor rigidity.
What you can do after
Apply it via molecular docking to rank compounds, inspect poses, recognize unreliable scores, and select candidates for stronger validation.
How it connects
Connects structure-based drug design with molecular docking, molecular dynamics, free-energy perturbation, and sequence-based virtual screening.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.