See the binding site. Understand it. Design against it.
Catalophore Studio turns a raw protein structure into an interpreted binding site in seconds. It unites point-cloud cavity detection, 30+ physicochemical property fields, structure-aware AI reasoning, ligand preparation, docking and proteome-wide cavity search, all in one interactive workbench, powered by Innophore’s Catalophore™ technology.
The workbench. A detected cavity rendered as a property point cloud, with structure hierarchy, sequence track, per-property histogram and tunnel detection in a single view.
One continuous path from a PDB to a design decision.
Catalophore Studio covers the whole path in one workbench. Each stage hands the next exactly what it needs: load a structure, prepare it, find and read the pocket, then prepare, dock and refine a ligand against it.
Structure
Open any PDB, fetch by ID from RCSB or UniProt, add hydrogens and parameterize ligands.
Cavities
Point-cloud detection annotates every pocket with 30+ property fields and geometry.
AI reasoning
A structured biological read of the site: zones, mechanism, druggability, ligand strategy.
Dock & refine
Prepare ligands, dock, score interactions and get AI-guided modification suggestions.
Every pocket, described as a field of measurable properties.
Catalophore Studio applies the validated Caviar algorithm to detect binding sites on a 3D grid, then annotates each grid point with the physicochemical environment a ligand would actually feel. The result is a property point cloud you can read, color, slice and quantify.
Shape you can trust
Volume, dimensions, sphericity, depth and lining residues for every detected cavity, plus tunnel and constriction detection for buried or channel-like sites.
30+ annotations per point
Hydrophobicity, polarity, nucleophilicity, Coulomb potential and electric field, element-specific van der Waals probes, metal / π–π / cation–π hotspots, flexibility and LigSite score.
Read it instantly
Color any property on the point cloud, inspect its distribution as a histogram, generate contour maps and titration profiles, and explore the structure in cartoon, stick or surface styles.
An expert read of the binding site, grounded in the numbers.
Catalophore Studio interprets the binding site, not just renders it. A deterministic analysis pipeline distills thousands of grid points into rigorous statistics, then a reasoning model returns a structured, evidence-cited biological assessment.
- Ab initio. The model sees only cavity geometry and properties, never the protein’s name, so the read is unbiased by prior annotation.
- Nine structured sections. Executive summary, geometry, composition, electrostatics, hydrophobicity pattern, functional zones, mechanism, EC classification and ligand recommendations.
- Choose the lens. Tell it whether your focus is catalysis or drug discovery and the interpretation adapts.
- Honest about itself. Every section reports a confidence level and explicit limitations, with no black-box certainty.
Structured AI interpretation. An executive summary with key features highlighted, plus a one-click choice of analysis lens.
Geometry & property statistics
Volume, sphericity and full descriptive statistics for all 30+ properties.
Spatial gradients & functional zones
Directional and radial trends; extrema clustered into catalytic centers, specificity pockets and exit channels.
Composition & residue influence
Lining residues classified by chemistry; each residue’s quantified effect on the local environment.
Reasoning & mechanism illustration
The curated report drives the structured interpretation and, for enzyme sites, a publication-style mechanism diagram.
A 10-module deterministic pipeline runs before any AI call, so interpretations are anchored in quantitative analysis rather than raw-data speculation.
A binding hypothesis is only as good as the chemistry of the molecule you test. AI structure and ligand pipelines move fast, but they still get stereochemistry wrong often enough to matter: by one benchmark, chirality errors occur in 4.4% of cases.
That is why Catalophore Studio builds ligand preparation on explicit, validated chemistry: RDKit-backed stereochemistry and tautomer analysis, stereoisomer and tautomer enumeration, plus GAFF2 atom typing and charge assignment via AnteCaviar. What you dock is chemically correct, not merely plausible.
Ligand preparation. Conformers, stereoisomers and tautomers enumerated with a 2D structure preview.
From a clean ligand to a scored, interpreted pose.
Once you understand the pocket, design against it. Catalophore Studio prepares ligands rigorously, docks them into the detected cavity, and analyses the result both geometrically and mechanistically.
- LigPrep. Enumerate conformers, stereoisomers and tautomers; analyze stereochemistry and bond orders before docking.
- AnteCaviar. GAFF2 atom typing and partial charges, exportable for downstream tools.
- Docking, single or batch. Pose generation, binding scores, full docking history and per-interaction breakdowns.
- SAR built in. R-group decomposition, matched molecular pairs, activity cliffs and scaffold analysis across a series.
The model tells you what to change, and why.
Every pose can be sent to an AI analysis that reads the interactions against the cavity’s own property fields, then proposes concrete, chemically reasoned modifications and a mechanistic view of likely PK behaviour.
- Modification suggestions tied to real features: fill volume near a constriction, rigidify a flexible chain to cut entropy loss, or add a basic donor to engage a specific aspartate.
- Mechanistic PK analysis that turns a binding result into a hypothesis about exposure and behaviour.
- Interaction mapping linking each suggestion back to hydrophobic, electrostatic or steric potential at the binding site.
Docking & mechanistic read. Scored poses with a one-click path into structure-aware AI analysis.
Actionable design advice. Concrete edits, each linked to the property driving it.
Find every pocket shaped like yours, across the human cavitome.
Catalophore Studio connects to Innophore’s CavitomiX / CATALObase, searching a precomputed library of human protein cavities by structural similarity rather than sequence. Two proteins that share no evolutionary ancestry can still have pockets shaped alike, and that is exactly what surfaces off-targets and repurposing opportunities.
Cavity, not sequence
Each cavity is encoded as an 80-dimensional embedding; cosine similarity ranks the closest matches across the cavitome, independent of fold or family.
See the liabilities early
Match a binding site proteome-wide and map hits back to UniProt, with duplicate filtering so each protein appears once, ranked by its best cavity.
New uses, known chemistry
The same search that flags off-targets surfaces proteins whose pockets match a known binder, a structural starting point for repurposing.
Sequence-based screening never sees a convergently evolved pocket. Cavity similarity does, and that is where both the risk and the opportunity hide.
The convergent-evolution principleEvery analysis becomes context the AI can reason over.
Catalophore Studio keeps a living knowledge base of your project. Cavity analyses, docking results and proteome matches are captured automatically; drop in a PDF paper and it is parsed for structures, database IDs and findings. Drag any entry into the chat and the copilot reasons with it.
Captured automatically
AI cavity reads, docking scores and top cavity matches are saved as structured knowledge entries the moment they complete.
PDFs, understood
Drop a paper to extract PDB / UniProt / ChEMBL / DrugBank / PubChem IDs, SMILES & InChI (RDKit-validated), activity tables and an AI summary.
Context-aware chat
Drag knowledge cards into the chat and ask questions; the assistant answers grounded in your own analyses and references.
The methods behind the workbench are published and proven.
Catalophore Studio is built on Innophore’s Catalophore™ point-cloud technology, a research line spanning a founding Nature Communications method, proteome-scale structure prediction with NVIDIA, and applications from drug repurposing to enzyme discovery.
Take stereochemistry into account when validating AI-generated structures
Nature 637, 548 (Correspondence) · why rigorous ligand chemistry matters
Folding the human proteome using BioNeMo: a fused dataset of structural models for machine learning
Nature Scientific Data 11, 591 · proteome-scale structural foundation, built with NVIDIA
CavitOmiX Drug Discovery: engineering antivirals with enhanced spectrum and reduced side effects
Viruses 16, 1186 · off-target reduction by cavity design
Fusidic & flufenamic acid identified as SARS-CoV-2 inhibitors via DrugSolver CavitomiX
Scientific Reports 13, 11783 · drug repurposing by cavity similarity
Identification of promiscuous ene-reductase activity by mining structural databases using active site constellations
Nature Communications 5 · the founding Catalophore™ proof-of-principle
Selected publications from the technology underlying Catalophore Studio.
Runs where your data lives.
A browser-based workbench backed by a documented REST API and an async compute layer. Deploy it locally, on a shared server, or on-prem GPU hardware, so your structures never have to leave your environment.
- API
Documented REST API
Programmatic structure submission, cavity detection, docking and job management.
- MCP
Model Context Protocol server
Search proteins, query the cavitome and launch analyses directly from AI assistants.
- GPU
On-prem & NVIDIA-ready
Containerized with all scientific binaries; cavitome inference runs on NVIDIA / DGX hardware.
- ⚡
Async & cached
Celery + Redis keep the UI responsive; results cache for instant re-analysis.
# detect cavities on a structure POST /api/caviar/detect { "pdb_id": "1ABC", "properties": "all" } # → ranked cavities + property fields { "cavities": [ { "id": 1, "volume": 2847, "points": 1247, "properties": "30+" } ] } # interpret a cavity with structured AI POST /api/cavity-reasoning/start { "pdb_id": "1ABC", "cavity_index": 1 }
Put the workbench to work.
Whether you run discovery programs or build the platforms behind them, there’s a way in.
Book a demo
Bring a target. We’ll detect its cavities, interpret the active site, and walk a lead series from preparation to AI-guided design, live.
Book a demo →Explore integration
Add point-cloud cavity analysis, AI interpretation and proteome-wide search to your environment via REST or MCP.
Talk integration →