Shape-based virtual screening

What is shape-based virtual screening?

Shape-based virtual screening is a computational method for ranking candidate molecules by how closely their three-dimensional conformers overlap a reference ligand or shape query. Methods may score molecular volume alone or combine shape with chemical features. The screen does not require a target structure, but its ranking depends on the chosen reference, candidate conformers, alignment method, and score.

Shape screening tests a different hypothesis from two-dimensional fingerprint similarity: molecules with different graphs may occupy a comparable volume in a bioactive conformation. Adding donors, acceptors, aromatic regions, or charge features can reject overlays that match the envelope but present incompatible chemistry. Neither variant confirms a shared target or mechanism.

The reference is often an experimentally observed bound ligand, but a carefully supported bioactive conformation can also be used. Flexible candidates require broader conformer ensembles, increasing compute and the chance of implausible high-scoring overlays. A defensible shortlist therefore retains aligned coordinates, conformer provenance, strain or geometry review, and scaffold diversity.

When to use shape-based virtual screening

Use a three-dimensional query when its conformation has a defensible biological basis:

• Use a bound ligand, experimentally supported pose, or carefully reviewed bioactive conformation as the shape query.
• Search for compounds that preserve a three-dimensional envelope or feature arrangement despite having a different molecular graph.
• Rank compounds from ligand information when receptor-based docking cannot be justified.

Strengths and limitations

A reviewed bioactive ligand can provide the query even when the receptor structure or binding site is uncertain. Three-dimensional overlap can identify candidates that preserve an overall shape or feature pattern without sharing the same two-dimensional substructure. Shape comparisons can rank large prepared libraries before slower docking, simulation, or experimental follow-up. Overlaying candidates with the reference makes the basis of a ranking easier to review than an unexplained aggregate score.

An irrelevant or strained reference pose can favor the wrong molecular envelope and exclude useful candidates. A compound cannot receive a strong overlap score if the prepared conformer ensemble does not represent the geometry needed for alignment. Two molecules can occupy similar volumes while presenting different electrostatic or interaction features. Large conformer ensembles increase compute and can produce high-scoring overlays that depend on unlikely conformations. A shape match is a prioritization signal, not evidence of target binding, activity, selectivity, safety, or efficacy.

Shape-only, feature-aware, and multi-reference screening

The query design controls what “similar” means. Validate each strategy on chemistry that resembles the intended library rather than assuming the most detailed score is automatically the most useful.

Shape-only overlap

Prioritize candidates that reproduce the reference volume when chemical-feature constraints would be premature or too restrictive.

Shape and feature overlap

Combine volume with compatible donors, acceptors, aromatic regions, charge centers, or other typed features.

Multi-reference screening

Use several justified ligands or conformations and preserve reference-specific ranks instead of hiding them behind an unexplained aggregate.

Shape-based virtual screening applications

Shape-based screening is useful for scaffold hopping, expanding around a bioactive ligand, and triaging libraries when a receptor structure is unavailable or unreliable. It can also provide a fast ligand-derived filter before a focused structure-based calculation.

A high overlap is least informative when the reference conformation is uncertain, candidates are highly flexible, or activity depends on directional interactions that the score omits. Review actual overlays and chemical features rather than selecting a shortlist from score values alone.

How to do shape-based virtual screening online

ProteinIQ currently provides a review workflow for aligned output from a dedicated shape-screening engine. The core overlay search remains external and should stay named in the run record.

1. Choose the reference conformation. Record the ligand source, biological context, stereochemistry, protonation state, and reason this geometry represents the query.
2. Standardize candidate structures. Preserve identifiers while resolving salts, stereochemistry, protonation, tautomers, duplicates, and invalid molecules.
3. Generate suitable conformers. Select conformer limits and energy settings that recover relevant geometries for validation compounds without uncontrolled ensemble growth.
4. Run the external overlay search. Apply the chosen shape-only or feature-aware method and export aligned SDF coordinates with reference identity, scores, and settings.
5. Review aligned matches in ProteinIQ. Check the aligned SDF structures with PoseBusters. Convert valid records with SDF to SMILES. Inspect properties from Molecular Descriptors. Review structural alerts from PAINS filter. Read endpoint estimates from ADMET-AI separately. Retain rejected and failed candidates.
6. Validate a diverse shortlist. Use an orthogonal target-aware method when available and test selected compounds in assays suited to the biological question.

How to interpret shape-screening results

Compare candidates within the same reference, conformer-generation policy, alignment method, and score definition. A score change after altering any of those inputs does not necessarily represent a meaningful change in biological plausibility.

Inspect whether the chosen overlay uses a strained or unusual conformer and whether key chemical features point in compatible directions. The ProteinIQ workflow reviews externally aligned hits; it does not recalculate their shape score. A diverse shortlist should represent several credible alignment hypotheses rather than many near-duplicate analogs of the top score. Similarity does not confirm binding, activity, safety or function.

How shape-based virtual screening works

A shape-based screen is a conformer comparison problem. Reference selection, candidate conformer generation, alignment, scoring, and diversity review all influence which compounds reach the shortlist.

1. Choose the reference shape. Select a relevant ligand conformation and document its source, protonation state, stereochemistry, and biological context.
2. Prepare candidate conformers. Standardize the candidate library, preserve identifiers, and generate a conformer ensemble that reflects the flexibility of each molecule.
3. Align and score shapes. Overlay candidate conformers with the reference and rank them using the chosen shape and chemical-feature scoring method.
4. Review chemistry and diversity. Inspect high-ranking overlays alongside molecular properties, liabilities, scaffold diversity, and unrealistic conformations.
5. Shortlist and validate. Export a diverse set for orthogonal computational review and experiments suited to the target and assay question.

Inputs and outputs

Check formats before running, then inspect and download the result from every workflow step.

Inputs

Reference ligand

SDFMOL2SMILES

Use a reviewed bioactive conformation when possible. Record its stereochemistry, protonation state, source, and biological context.

Candidate library

SMILESSDF

Provide standardized structures with stable identifiers so conformers, scores, and exclusions can be traced back to each compound.

Shape and feature settings

Define conformer limits, alignment behavior, shape-versus-feature weighting, and any score or diversity thresholds before screening.

Outputs

Shape-ranked compounds

CSVJSON

Retain the best score, selected conformer, reference identity, and method settings for every ranked candidate.

Aligned conformers

SDFMOL2

Inspect and export candidate overlays in the coordinate frame of the reference query.

Property and diversity review

CSVJSON

Compare molecular properties, alerts, scaffold groups, and selection decisions alongside the shape ranking.

Run record

LOGFILES

Keep preparation failures, conformer settings, ranking parameters, and exported files connected to the screen.