Skills
skills/mims-harvard/tooluniverse/tooluniverse-binder-discovery

tooluniverse-binder-discovery

SKILL.md

Small Molecule Binder Discovery Strategy

Systematic discovery of novel small molecule binders using 60+ ToolUniverse tools across druggability assessment, known ligand mining, similarity expansion, ADMET filtering, and synthesis feasibility.

KEY PRINCIPLES:

  1. Report-first approach - Create report file FIRST, then populate progressively
  2. Target validation FIRST - Confirm druggability before compound searching
  3. Multi-strategy approach - Combine structure-based and ligand-based methods
  4. ADMET-aware filtering - Eliminate poor compounds early
  5. Evidence grading - Grade candidates by supporting evidence
  6. Actionable output - Provide prioritized candidates with rationale
  7. English-first queries - Always use English terms in tool calls, even if the user writes in another language. Only try original-language terms as a fallback. Respond in the user's language

Critical Workflow Requirements

1. Report-First Approach (MANDATORY)

DO NOT show search process or tool outputs to the user. Instead:

  1. Create the report file FIRST - Before any data collection:

    • File name: [TARGET]_binder_discovery_report.md
    • Initialize with all section headers from the template (see REPORT_TEMPLATE.md)
    • Add placeholder text: [Researching...] in each section

  2. Progressively update the report - As you gather data:

    • Update each section with findings immediately
    • The user sees the report growing, not the search process

  3. Output separate data files:

    • [TARGET]_candidate_compounds.csv - Prioritized compounds with SMILES, scores
    • [TARGET]_bibliography.json - Literature references (optional)

2. Citation Requirements (MANDATORY)

Every piece of information MUST include its source:

*Source: ChEMBL via `ChEMBL_get_target_activities` (CHEMBL203)*
*Source: PDB via `get_protein_metadata_by_pdb_id` (1M17)*
*Source: ADMET-AI via `ADMETAI_predict_toxicity`*
*Source: NVIDIA NIM via `NvidiaNIM_alphafold2` (pLDDT: 90.94)*

Workflow Overview

Phase 0: Tool Verification (check parameter names)
    |
Phase 1: Target Validation
    |- 1.1 Resolve identifiers (UniProt, Ensembl, ChEMBL target ID)
    |- 1.2 Assess druggability/tractability
    |   +- 1.2a GPCRdb integration (for GPCR targets)
    |   +- 1.2.5 Check therapeutic antibodies (Thera-SAbDab)
    |- 1.3 Identify binding sites
    +- 1.4 Predict structure (NvidiaNIM_alphafold2/esmfold)
    |
Phase 2: Known Ligand Mining
    |- ChEMBL bioactivity data
    |- GtoPdb interactions
    |- Chemical probes (Open Targets)
    |- BindingDB affinity data (Ki/IC50/Kd)
    |- PubChem BioAssay HTS data (screening hits)
    +- SAR analysis from known actives
    |
Phase 3: Structure Analysis
    |- PDB structures with ligands
    |- EMDB cryo-EM structures (for membrane targets)
    |- Binding pocket analysis
    +- Key interactions
    |
Phase 3.5: Docking Validation (NvidiaNIM_diffdock/boltz2)
    |- Dock reference inhibitor
    +- Validate binding pocket geometry
    |
Phase 4: Compound Expansion
    |- 4.1-4.3 Similarity/substructure search
    +- 4.4 De novo generation (NvidiaNIM_genmol/molmim)
    |
Phase 5: ADMET Filtering
    |- Physicochemical properties (Lipinski, QED)
    |- Bioavailability, toxicity, CYP interactions
    +- Structural alerts (PAINS)
    |
Phase 6: Candidate Docking & Prioritization
    |- Dock all candidates (NvidiaNIM_diffdock/boltz2)
    |- Score by docking (40%) + ADMET (30%) + similarity (20%) + novelty (10%)
    |- Assess synthesis feasibility
    +- Generate final ranked list (top 20)
    |
Phase 6.5: Literature Evidence
    |- PubMed (peer-reviewed SAR studies)
    |- EuropePMC preprints (source='PPR')
    +- OpenAlex citation analysis
    |
Phase 7: Report Synthesis & Delivery

Phase 0: Tool Verification

CRITICAL: Verify tool parameters before calling unfamiliar tools.

tool_info = tu.tools.get_tool_info(tool_name="ChEMBL_get_target_activities")

Known Parameter Corrections

Tool WRONG Parameter CORRECT Parameter
OpenTargets_* ensembl_id ensemblId (camelCase)
ChEMBL_get_target_activities chembl_target_id target_chembl_id
ChEMBL_search_similar_molecules smiles molecule (accepts SMILES, ChEMBL ID, or name)
alphafold_get_prediction uniprot accession
ADMETAI_* smiles="..." smiles=["..."] (must be list)
NvidiaNIM_alphafold2 seq sequence
NvidiaNIM_genmol smiles="C..." smiles="C...[*{1-3}]..." (must have mask)
NvidiaNIM_boltz2 sequence="..." polymers=[{"molecule_type": "protein", "sequence": "..."}]

Phase 1: Target Validation

1.1 Identifier Resolution

Resolve all IDs upfront and store for downstream queries:

1. UniProt_search(query=target_name, organism="human") -> UniProt accession
2. MyGene_query_genes(q=gene_symbol, species="human") -> Ensembl gene ID
3. ChEMBL_search_targets(query=target_name, organism="Homo sapiens") -> ChEMBL target ID
4. GtoPdb_get_targets(query=target_name) -> GtoPdb ID (if GPCR/channel/enzyme)

1.2 Druggability Assessment

Use multi-source triangulation:

  • OpenTargets_get_target_tractability_by_ensemblID(ensemblId) - tractability bucket
  • DGIdb_get_gene_druggability(genes=[gene_symbol]) - druggability categories
  • OpenTargets_get_target_classes_by_ensemblID(ensemblId) - target class
  • For GPCRs: GPCRdb_get_protein + GPCRdb_get_ligands + GPCRdb_get_structures
  • For antibody landscape: TheraSAbDab_search_by_target(target=target_name)

Decision Point: If druggability < 2 stars, warn user about challenges.

1.3 Binding Site Analysis

  • ChEMBL_search_binding_sites(target_chembl_id)
  • get_binding_affinity_by_pdb_id(pdb_id) for co-crystallized ligands
  • InterPro_get_protein_domains(accession) for domain architecture

1.4 Structure Prediction (NVIDIA NIM)

Requires NVIDIA_API_KEY. Two options:

  • AlphaFold2: NvidiaNIM_alphafold2(sequence, algorithm="mmseqs2") - high accuracy, 5-15 min
  • ESMFold: NvidiaNIM_esmfold(sequence) - fast (~30s), max 1024 AA

Always report pLDDT confidence scores (>=90 very high, 70-90 confident, <70 caution).

Phase 2: Known Ligand Mining

Tools (in order of priority)

Source Tool Strengths
ChEMBL ChEMBL_get_target_activities Curated, SAR-ready
BindingDB BindingDB_get_ligands_by_uniprot Direct Ki/Kd, literature links
GtoPdb GtoPdb_get_target_interactions Pharmacology focus (GPCRs, channels)
PubChem PubChem_search_assays_by_target_gene HTS screens, novel scaffolds
Open Targets OpenTargets_get_chemical_probes_by_target_ensemblID Validated probes

Key Steps

  1. Get all bioactivities: filter to IC50/Ki/Kd < 10 uM
  2. Get molecule details for top actives: ChEMBL_get_molecule
  3. Identify chemical probes and approved drugs
  4. Analyze SAR: common scaffolds, key modifications
  5. Check off-target selectivity: BindingDB_get_targets_by_compound

Phase 3: Structure Analysis

Tools

  • PDB_search_similar_structures(query=uniprot, type="sequence") - find PDB entries
  • get_protein_metadata_by_pdb_id(pdb_id) - resolution, method
  • get_binding_affinity_by_pdb_id(pdb_id) - co-crystal ligand affinities
  • get_ligand_smiles_by_chem_comp_id(chem_comp_id) - ligand SMILES from PDB
  • emdb_search(query) - cryo-EM structures (prefer for GPCRs, ion channels)
  • alphafold_get_prediction(accession) - AlphaFold DB fallback

Phase 3.5: Docking Validation (NVIDIA NIM)

Situation Tool Input
Have PDB + SDF NvidiaNIM_diffdock protein=PDB, ligand=SDF, num_poses=10
Have sequence + SMILES NvidiaNIM_boltz2 polymers=[...], ligands=[...]

Dock a known reference inhibitor first to validate the binding pocket.

Phase 4: Compound Expansion

4.1-4.3 Search-Based Expansion

Use 3-5 diverse actives as seeds, similarity threshold 70-85%:

  • ChEMBL_search_similar_molecules(molecule=SMILES, similarity=70)
  • PubChem_search_compounds_by_similarity(smiles, threshold=0.7)
  • ChEMBL_search_substructure(smiles=core_scaffold)
  • STITCH_get_chemical_protein_interactions(identifier=gene, species=9606)

4.4 De Novo Generation (NVIDIA NIM)

GenMol - scaffold hopping with masked regions:

NvidiaNIM_genmol(smiles="...core...[*{3-8}]...tail...[*{1-3}]...", num_molecules=100, temperature=2.0, scoring="QED")

Mask syntax: [*{min-max}] specifies atom count range.

MolMIM - controlled analog generation:

NvidiaNIM_molmim(smi=reference_smiles, num_molecules=50, algorithm="CMA-ES")

Phase 5: ADMET Filtering

Apply filters sequentially (all take smiles=[list]):

Step Tool Filter Criteria
Physicochemical ADMETAI_predict_physicochemical_properties Lipinski <= 1, QED > 0.3, MW 200-600
Bioavailability ADMETAI_predict_bioavailability Oral bioavailability > 0.3
Toxicity ADMETAI_predict_toxicity AMES < 0.5, hERG < 0.5, DILI < 0.5
CYP ADMETAI_predict_CYP_interactions Flag CYP3A4 inhibitors
Alerts ChEMBL_search_compound_structural_alerts No PAINS

Include a filter funnel table in the report showing pass/fail counts at each stage.

Phase 6: Candidate Docking & Prioritization

Scoring Framework

Dimension Weight Source
Docking confidence 40% NvidiaNIM_diffdock/boltz2
ADMET score 30% ADMETAI predictions
Similarity to known active 20% Tanimoto coefficient
Novelty 10% Not in ChEMBL + novel scaffold bonus

Evidence Tiers

Tier Criteria
T0 (4 stars) Docking score > reference inhibitor
T1 (3 stars) Experimental IC50/Ki < 100 nM
T2 (2 stars) Docking within 5% of reference OR IC50 100-1000 nM
T3 (1 star) >80% similarity to T1 compound
T4 (0 stars) 70-80% similarity, scaffold match
T5 (empty) Generated molecule, ADMET-passed, no docking

Deliver top 20 candidates with: Rank, ID, SMILES, Docking score, ADMET score, overall score, source, evidence tier.

Phase 6.5: Literature Evidence

  • PubMed_search_articles(query="[TARGET] inhibitor SAR") - peer-reviewed
  • EuropePMC_search_articles(query, source="PPR") - preprints (not peer-reviewed)
  • openalex_search_works(query) - citation analysis

Fallback Chains

Target ID:     ChEMBL_search_targets -> GtoPdb_get_targets -> "Not in databases"
Druggability:  OpenTargets tractability -> DGIdb druggability -> target class proxy
Bioactivity:   ChEMBL -> BindingDB -> GtoPdb -> PubChem BioAssay -> "No data"
Structure:     PDB -> EMDB (membrane) -> NvidiaNIM_alphafold2 -> NvidiaNIM_esmfold -> AlphaFold DB -> "None"
Similarity:    ChEMBL similar -> PubChem similar -> "Search failed"
Docking:       NvidiaNIM_diffdock -> NvidiaNIM_boltz2 -> similarity-based scoring
Generation:    NvidiaNIM_genmol -> NvidiaNIM_molmim -> similarity search only
Literature:    PubMed -> EuropePMC (preprints) -> OpenAlex
GPCR data:     GPCRdb_get_protein -> GtoPdb_get_targets

NVIDIA NIM Runtime Reference

Tool Runtime Notes
NvidiaNIM_alphafold2 5-15 min Async, max ~2000 AA
NvidiaNIM_esmfold ~30 sec Max 1024 AA
NvidiaNIM_diffdock ~1-2 min Per ligand
NvidiaNIM_boltz2 ~2-5 min End-to-end complex
NvidiaNIM_genmol ~1-3 min Depends on num_molecules
NvidiaNIM_molmim ~1-2 min Close analog generation

Always check: import os; nvidia_available = bool(os.environ.get("NVIDIA_API_KEY"))

Rate Limiting

Database Limit Strategy
ChEMBL ~10 req/sec Batch queries
PubChem ~5 req/sec Batch endpoints
ADMET-AI No strict limit Batch SMILES in lists
NVIDIA NIM API key quota Cache results

For large expansions (>500 compounds): batch in chunks of 100, prioritize top candidates for docking.

Reference Files

For detailed protocols, examples, and templates, see:

File Contents
WORKFLOW_DETAILS.md Phase-by-phase procedures, code patterns, screening protocols, fallback chain details
TOOLS_REFERENCE.md Complete tool reference with parameters, usage examples, and fallback chains
REPORT_TEMPLATE.md Report file template, evidence grading system, section formatting examples
EXAMPLES.md End-to-end workflow examples (EGFR, novel target, lead optimization, NVIDIA NIM)
CHECKLIST.md Pre-delivery verification checklist for report quality
Weekly Installs
136
Repository
mims-harvard/to…universe
GitHub Stars
1.1K
First Seen
Feb 7, 2026
Security Audits
Gen Agent Trust HubPass
SocketPass
SnykWarn
Installed on
gemini-cli128
opencode127
codex127
github-copilot125
amp120
kimi-cli120