Establish biological context for a drug target protein.

Retrieves protein function, family classification, subcellular location, disease associations, and AlphaFold structure confidence. This should be your FIRST call — all downstream analysis depends on this context.

IMPORTANT: AlphaFold confidence flags regions where predicted structure is unreliable. Low-confidence regions (<70 pLDDT) may have incorrect pocket predictions — always cross-reference with experimental structures.

Provide EITHER a PDB ID (e.g. '1M17') OR a UniProt accession (e.g. 'P00533'). If a PDB ID is given, the UniProt mapping is resolved automatically.

Find all PDB structures for a target protein.

Returns all available experimental structures, sorted by resolution. Use this to understand structural coverage: how many structures exist, what ligands have been co-crystallized, what conformational states are captured, and what the best-quality structure is.

A target with many high-resolution co-crystal structures has a rich binding site landscape to analyze. A target with only 1-2 structures (or only apo/unliganded structures) has less structural evidence.

Call this AFTER CharacterizeTarget and BEFORE GetBindingSites to identify which structures to analyze for pockets.

Map all known binding sites in a protein structure from co-crystallized ligands.

Identifies binding pockets by analyzing non-polymer entities (ligands, cofactors) in the structure, filtering out crystallization artifacts (glycerol, PEG, sulfate, etc.), and classifying each site by type.

Each site includes druggability assessment and modality recommendations:

• Small deep pockets (< 20 contact residues) favor small molecules

• Large flat interfaces (> 30 residues) favor biologics or de novo protein binders

• Allosteric sites may offer selectivity advantages over orthosteric sites

Call this on specific PDB IDs identified by GetRelatedStructures. For comprehensive analysis, call on multiple structures with different co-crystallized ligands to build a complete pocket map.

Retrieve the bioactivity landscape for a drug target from ChEMBL.

Shows what compounds have been tested, how potent the best ones are, whether any have reached clinical trials, and how crowded the competitive landscape is.

Use this to decide whether to target KNOWN binding sites (where SAR exists) or seek NOVEL sites (where de novo design has an advantage). A crowded landscape suggests new modalities or allosteric approaches; an untargeted landscape suggests opportunity but less prior validation.

Provide EITHER uniprot_id or pdb_id (UniProt preferred for accuracy).

Check conservation at binding site residues across mouse, rat, and cynomolgus.

Critical for preclinical translatability: if key binding site residues differ between human and a preclinical model, that species' efficacy data may not predict human response. Non-conserved positions are flagged for each species individually.

Conservation > 90%: excellent — species should recapitulate human binding. Conservation 70-90%: acceptable — verify non-conserved positions are not critical contact residues. Conservation < 70%: caution for that species — consider a better-conserved alternative. Cynomolgus (macaque) is the closest primate model and often shows higher conservation than rodents when the target has primate-specific sequence features.

Default species checked: mouse, rat, cynomolgus. Pass a custom species list to restrict or reorder (supported values: 'mouse', 'rat', 'cynomolgus').

Provide the human UniProt accession and residue positions from GetBindingSites.

Flag known sequence variants and mutagenesis hits at binding-site residues.

Binding-site residues that are documented disease/resistance variants (e.g. EGFR T790M) mark pockets that mutate under drug pressure — a key risk signal when choosing where to design. Source: UniProt Natural variant + Mutagenesis features. Provide the human UniProt accession and positions from GetBindingSites.

Search PubMed for recent structural biology and drug design papers.

Focuses specifically on binding site characterization, allosteric mechanisms, resistance mutations, and prior design campaigns — the literature most relevant to planning a new binder design effort.

Use the optional context parameter to narrow results, e.g.:

• context='allosteric' for allosteric site literature

• context='resistance' for resistance mutation papers

• context='antibody' for biologic-focused papers

• context='oncology' for disease-specific context

Call this LAST — after CharacterizeTarget, GetRelatedStructures, GetBindingSites, GetLigandHistory, and CheckConservation — to see if the literature reveals insights not captured in database records (e.g., cryptic sites found by MD simulation, unpublished allosteric mechanisms).

Map the union of binding pockets across all structures of a target.

Fans out GetBindingSites over the top structures and clusters pockets by residue overlap, so recurrent pockets (e.g. the ATP site appearing in most structures) stand out from one-off or artifact sites. The heaviest tool — downloads several coordinate files. Provide uniprot_id (preferred) or pdb_id.