name: bio-shape-similarity description: Performs 3D shape-based similarity searching using ROCS (OpenEye), USRCAT (ultra-fast), Open3DAlign (RDKit), ESPSim (electrostatic), and ShaEP with explicit handling of Tanimoto-Combo (shape + color), shape vs ECFP4 complementarity, conformer-ensemble searching, alignment optimization, and scaffold hopping. Use when searching for shape-mimicking compounds with different scaffolds, identifying bioisosteric replacements, prospective scaffold hopping, or expanding hit series beyond 2D similarity. origin: openai4s category: bioskills/chemoinformatics metadata: tool_type: python primary_tool: RDKit third_party: name: GPTomics/bioSkills repository: https://github.com/GPTomics/bioSkills commit: d91ed3d563019e649dc854c56ccd62551359488a license: MIT
Reference examples tested with: RDKit 2024.09+ (Open3DAlign and USRCAT); official ShaEP syntax checked against ShaEP 1.4.2; ROCS/FastROCS/ROCS X are commercial OpenEye products.
Before using code patterns, verify installed versions match. If versions differ:
- Python: pip show <package> then help(module.function) to check signatures
If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.
Search for compounds with similar 3D shape (and optionally chemical features) to a query molecule. Shape-based screening complements 2D fingerprint search: it can find scaffold-hopped compounds that ECFP4 misses (different scaffolds with similar shape). ROCS (OpenEye) is the industry-standard commercial tool; Open3DAlign (RDKit), USRCAT (Schreyer & Blundell 2012), and ShaEP are open-source alternatives. Modern best practice combines shape with color (chemical-feature similarity) via Tanimoto-Combo: matches share both shape and pharmacophore feature distribution.
For 2D fingerprint similarity, see chemoinformatics/similarity-searching. For pharmacophore search (discrete feature constraints), see chemoinformatics/pharmacophore-modeling. For 3D conformer generation, see chemoinformatics/conformer-generation.
| Tool | Speed | Approach | Open-source | Fails when |
|---|---|---|---|---|
| ROCS / FastROCS (OpenEye) | Hardware/database/conformer-dependent; vendor reports millions of conformers/s for FastROCS | Gaussian shape + color | No | License and prepared database |
| ROCS X | Trillion-scale reaction/synthon space on Orion | FastROCS plus Bayesian-bandit sampling | No | Commercial cloud workflow |
| USRCAT | Very fast alignment-free descriptor comparison | Moment-based + atom types | Yes | Coarse approximation |
| Open3DAlign (RDKit) | Medium | MMFF atom-type/charge-weighted alignment | Yes | Requires compatible typed 3D structures |
| ShaEP | Benchmark on actual conformers/hardware | Field-based (shape + ESP) | Free binary; inspect license | Requires valid 3D structures and charges for ESP |
| ESPSim | Benchmark on actual workload | Electrostatic + shape | Yes | Limited public benchmarks |
| Phase-Shape (Schrödinger) | commercial | Shape + pharmacophore | No | Commercial |
| USR (original) | Very fast alignment-free comparison | Moment-based only | Yes | No atom-type information |
Decision: Select a shape method by matched retrieval/enrichment performance, conformer preparation, throughput, licensing, and score semantics. USRCAT is useful as a fast prefilter; Open3DAlign provides an open alignment method; ROCS/FastROCS provide commercial shape/color workflows.
| Scenario | Method | Notes |
|---|---|---|
| Large prepared library | USRCAT pre-filter + Open3DAlign rescore | Choose rescore budget from measured retrieval saturation |
| Production VS for scaffold hop | ROCS + color (commercial) | Industry standard |
| Scaffold hopping prospective | Open3DAlign with conformer ensemble | Shape + flexibility |
| Bioisostere replacement | ROCS color with neutral scoring | Pharmacophore-equivalent matches |
| Patent space carve-out | Shape constraint + 2D dissimilarity | Combine shape + dissimilar scaffold |
| Library diversity assessment | USRCAT k-nearest neighbor | Fast |
| Crystal-bound conformer template | Open3DAlign starting from co-crystal pose | Bioactive shape |
| Cross-target screening | Shape + pharmacophore feature | Combined screen |
TanimotoCombo = Tanimoto_shape + Tanimoto_color
Each component is normalized from 0 to 1, so TanimotoCombo ranges from 0 to 2. It is a sum, not an average. Select follow-up thresholds from a relevant benchmark or enrichment study; a single cutoff is not portable across query preparation, color-force-field settings, and library composition.
USRCAT (Schreyer & Blundell 2012) extends Ultrafast Shape Recognition (USR) with atom-type information. Each molecule is represented as a 60-dimensional moment vector (12 moments × 5 atom types).
Goal: Encode a molecule into the 60-D USRCAT moment vector and score similarity against another molecule for alignment-free shape search.
Approach: Parse the SMILES, add hydrogens, generate one 3D conformer with ETKDGv3, compute RDKit USRCAT descriptors, and compare descriptor vectors with RDKit's USR score.
from rdkit.Chem import rdMolDescriptors
mol = Chem.MolFromSmiles('CCO')
mol = Chem.AddHs(mol)
AllChem.EmbedMolecule(mol, AllChem.ETKDGv3())
descriptors = rdMolDescriptors.GetUSRCAT(mol)
# Returns numpy array of 60 floats: 12 USR moments x 5 atom types
# (all atoms, hydrophobic, aromatic, acceptor, donor)
similarity = rdMolDescriptors.GetUSRScore(desc1, desc2)
Speed: Descriptor calculation is linear in atoms and comparison is fixed-length, without pairwise alignment. Benchmark end-to-end throughput on the prepared conformer library before choosing a scale cutoff.
Limit: USRCAT is a coarse approximation. Predictive for analog identification; less precise for scaffold hopping.
Open3DAlign uses MMFF atom types and partial charges to find an atom-based 3D alignment:
Goal: Align a target molecule onto a query in 3D and score volume overlap with Open3DAlign.
Approach: Build 3D structures for query and target, run GetO3A, and call Align() to transform the probe in place. Score() is the unnormalized O3A objective, not a shape Tanimoto or ROCS TanimotoCombo. If a normalized shape similarity is required, compute 1 - rdShapeHelpers.ShapeTanimotoDist(...) after alignment.
from rdkit.Chem import rdMolAlign, rdShapeHelpers
query = Chem.MolFromSmiles('CCC(=O)Nc1ccccc1')
query = Chem.AddHs(query)
AllChem.EmbedMolecule(query, AllChem.ETKDGv3())
target = Chem.MolFromSmiles('CCC(=O)Nc1ccc(F)cc1')
target = Chem.AddHs(target)
AllChem.EmbedMolecule(target, AllChem.ETKDGv3())
O3A = rdMolAlign.GetO3A(target, query)
rmsd = O3A.Align() # aligns target to query in place
o3a_score = O3A.Score()
shape_tanimoto = 1.0 - rdShapeHelpers.ShapeTanimotoDist(target, query)
GetO3A finds an alignment between conformers; Align() applies it and returns RMSD. Keep o3a_score and normalized shape_tanimoto distinct in outputs.
Open3DAlign vs ROCS: Open3DAlign is open-source and competitive on small benchmarks; slower than ROCS at scale.
For each library molecule, generate ensemble of conformers; pick best-shape conformer:
Goal: Run shape-similarity search over a conformer ensemble per library molecule so bound-conformer-like shapes are recovered.
Approach: For each library molecule, add hydrogens, embed n_conf conformers with ETKDGv3, MMFF-optimize, score each conformer against the query with Open3DAlign, and keep the best score per molecule.
def shape_search_ensemble(query_mol, library_mols, n_conf=20):
hits = []
for target in library_mols:
target = Chem.AddHs(target)
ids = list(AllChem.EmbedMultipleConfs(target, numConfs=n_conf,
params=AllChem.ETKDGv3()))
if not ids:
continue
if not AllChem.MMFFHasAllMoleculeParams(target):
continue
optimization = AllChem.MMFFOptimizeMoleculeConfs(target)
if any(status != 0 for status, _ in optimization):
continue
scores = []
for c in range(target.GetNumConformers()):
O3A = rdMolAlign.GetO3A(target, query_mol, prbCid=c)
O3A.Align()
scores.append(1.0 - rdShapeHelpers.ShapeTanimotoDist(
target, query_mol, confId1=c,
))
if scores:
hits.append((target, max(scores)))
return sorted(hits, key=lambda x: x[1], reverse=True)
Critical: Results depend on conformer coverage. Use an ensemble sized and validated for the library and query rather than assuming one conformer is representative.
ShaEP and ESPSim extend shape with electrostatic surface potential overlap. For ESP-relevant pharmacophores (binding pockets with strong electrostatics):
shaep -q query.mol2 target.mol2 -s aligned_hits.sdf similarity.txt
ESP scoring catches electrostatic-equivalent bioisosteres that pure shape misses (carboxylate vs tetrazole same charge).
| Shape result | ECFP4 result | Interpretation |
|---|---|---|
| High | High | Close analog candidate |
| High | Low | Scaffold-hop candidate |
| Low | High | Similar 2D chemotype in a different sampled shape |
| Low | Low | Unrelated by these representations |
Calibrate “high” and “low” on a task-relevant reference set; do not treat the illustrative function defaults below as universal scientific cutoffs.
The shape >> ECFP4 quadrant is the scaffold-hopping gold:
Goal: Identify scaffold-hop candidates that are 3D-shape-similar but 2D-chemotype-dissimilar to the query.
Approach: Run the conformer-ensemble shape search, keep hits above a shape Tanimoto cutoff, then retain only those whose ECFP4 Tanimoto to the query is below an ECFP4 dissimilarity cutoff.
# These thresholds are repository starting defaults only; calibrate both on a
# task-relevant active/decoy or retrieval benchmark before making decisions.
def scaffold_hop_candidates(query_mol, library, shape_threshold=0.7,
ecfp_threshold=0.5):
shape_hits = shape_search_ensemble(query_mol, library)
candidates = []
for target, shape_score in shape_hits:
if shape_score >= shape_threshold:
ecfp_sim = ecfp_tanimoto(query_mol, target)
if ecfp_sim < ecfp_threshold:
candidates.append((target, shape_score, ecfp_sim))
return candidates
Trigger: Library has many fragment-sized compounds.
Mechanism: USRCAT moments dominated by overall shape; small molecules look "similar" if shape resemble.
Symptom: Many fragment hits; not pharmacophore-relevant.
Fix: Calibrate size/property filters on the retrieval task and rescore selected hits with an alignment or feature-aware method.
Trigger: Million-compound library, full alignment.
Mechanism: Open3DAlign is iterative; O(N) per molecule.
Symptom: Hours of compute.
Fix: Pre-filter with USRCAT and choose the rescore budget from measured throughput and retrieval saturation.
Trigger: Mirror-image of correct binder.
Mechanism: Shape-only scoring may insufficiently penalize stereochemical alternatives even though a rigid rotational overlay is not generally invariant to mirror reflection.
Symptom: Enantiomer of inactive scores as hit.
Fix: Validate hits by 3D pose; check stereochemistry.
Trigger: -COOH replaced by -SO3H or tetrazole.
Mechanism: Default color types may not equate these bioisosteres.
Symptom: Known bioisostere doesn't score high.
Fix: Validate the color-force-field treatment for the bioisostere and compare shape, color, and pharmacophore evidence separately.
Trigger: Library compound generated conformer is not the bound conformation.
Mechanism: ETKDGv3 generates plausible conformers; bound conformer may be higher energy.
Symptom: Known active doesn't shape-match query.
Fix: Use larger conformer ensemble; weight by Boltzmann; or use CREST + GFN2-xTB for high-quality sampling.
Trigger: ShaEP or ESPSim on production library.
Mechanism: Field-based methods compute Gaussian fields per molecule.
Symptom: Field calculation or alignment dominates runtime on the prepared library.
Fix: Use as second-stage rescore; not primary screen.
| Aspect | Shape | Pharmacophore |
|---|---|---|
| Representation | Volume distribution | Discrete features in space |
| Captures | Overall bulk | Interaction-relevant features |
| Speed | Fast (USRCAT) to medium (Open3DAlign) | Fast |
| Specificity | Task- and query-dependent | Task- and feature-definition-dependent |
| False positive rate | Measure on a matched benchmark | Measure on a matched benchmark |
| Best for | Scaffold hopping initial | Scaffold hopping refinement |
Shape and pharmacophore searches make different approximations. Compare them alone and in sequence on a matched active/decoy or retrieval benchmark before assigning recall/precision roles.
| Symptom | Cause | Fix |
|---|---|---|
| Open3DAlign RMSD is near 0 | Near-exact O3A alignment | Treat as a successful alignment; evaluate the O3A and shape scores separately |
| USRCAT vector all zeros | Mol has no 3D coords | Generate conformer first |
| Shape Tanimoto > 1 | Raw O3A or TanimotoCombo mislabeled as shape Tanimoto | Shape Tanimoto is 0-1; O3A is unnormalized and ROCS TanimotoCombo is 0-2 |
| ROCS very slow | Sequential processing | Use parallel batching |
| Shape match but no docking pose | Wrong binding pose | Use docking on top shape hits, not shape alone |
| Missing co-crystal template | Apo or AlphaFold-only structure | Use ligand-based pharmacophore + shape |
| ShaEP returns no hits | Strict tolerance | Loosen overlap thresholds |