ChEMBL Connector#

The ChEMBL Connector gives Claude access to the ChEMBL Database, a manually curated resource of bioactive drug-like compounds with quantitative binding and functional data against biological targets.

What You Can Do#

With this connector, Claude can help you:

  • Search for compounds by name or chemical structure

  • Retrieve bioactivity data (IC50, EC50, Ki values)

  • Find drug targets and mechanisms of action

  • Access ADMET properties for drug candidates

  • Discover approved drugs by indication

  • Research structure-activity relationships

Data Source: ChEMBL Database v34 (EMBL-EBI)

MCP Endpoint: https://mcp.deepsense.ai/chembl/mcp Coverage: 2.4+ million compounds, 1.5+ million assays, 16+ million activities


Getting Started#

Adding the Connector to Claude#

  1. Open Claude’s Connector Settings

    • In Claude Desktop, select your profile icon in the bottom left

    • Navigate to Settings → Connectors

  2. Find the ChEMBL Connector

    • Browse the available connectors catalogue

    • Search for “ChEMBL” or find it under Life Sciences connectors

  3. Enable the Connector

    • Click on the ChEMBL Connector

    • Click “Enable” or “Add Connector”

    • The connector will be activated immediately

    • No authentication required - the connector works immediately after enabling

  4. Verify the Connection

    • Ask Claude: “What is the IC50 of imatinib against BCR-ABL?”

    • Claude should use the ChEMBL connector to find bioactivity data


Available Tools#



3. Get Bioactivity#

What it does: Retrieves bioactivity data for compound-target interactions.

Use it for:

  • Finding IC50, EC50, or Ki values

  • Comparing compound potencies

  • Assessing selectivity profiles

  • Identifying off-target activities

Example queries:

  • “What is the IC50 of imatinib against BCR-ABL?”

  • “Show me bioactivity data for aspirin”

  • “Find Ki values for gefitinib”

  • “What targets does pembrolizumab bind to?”

  • “Compare IC50 values of erlotinib and gefitinib”

What Claude will do:

  1. Retrieve activity data for the compound

  2. Show quantitative measurements:

    • IC50 (half maximal inhibitory concentration)

    • EC50 (half maximal effective concentration)

    • Ki (inhibition constant)

    • Kd (dissociation constant)

    • Values in nM (nanomolar) or μM (micromolar)

  3. Indicate target names and organisms

  4. Provide assay information and confidence scores

  5. Include data quality indicators


4. Get Mechanism#

What it does: Retrieves mechanism of action and target binding information.

Use it for:

  • Understanding how drugs work

  • Identifying primary and secondary targets

  • Finding action types (inhibitor, agonist, antagonist)

  • Researching polypharmacology

Example queries:

  • “How does imatinib work?”

  • “What is the mechanism of action of aspirin?”

  • “What targets does gefitinib inhibit?”

  • “Show me the mechanism for pembrolizumab”

What Claude will do:

  1. Retrieve mechanism of action descriptions

  2. Identify primary target(s)

  3. Show action type:

    • Inhibitor (decreases activity)

    • Agonist (activates target)

    • Antagonist (blocks activation)

    • Blocker, Modulator, Activator, etc.

  4. Indicate clinical development phase

  5. Provide references to publications



6. Get ADMET#

What it does: Retrieves ADMET properties (Absorption, Distribution, Metabolism, Excretion, Toxicity).

Use it for:

  • Predicting drug-likeness

  • Assessing safety liabilities

  • Comparing pharmacokinetic profiles

  • Lead optimization

  • Understanding metabolic stability

Example queries:

  • “What are the ADMET properties of imatinib?”

  • “Show me pharmacokinetic data for aspirin”

  • “Is gefitinib orally bioavailable?”

  • “What is the LogP of pembrolizumab?”

What Claude will do:

  1. Retrieve calculated molecular properties:

    • ALogP: Lipophilicity (optimal: 1-3 for oral drugs)

    • Molecular Weight: Full molecular weight in Da

    • PSA: Polar surface area (<140 for oral, <90 for CNS)

    • HBA/HBD: H-bond acceptors/donors (Rule of 5: ≤10/≤5)

    • Rotatable Bonds: <10 for good oral bioavailability

    • Aromatic Rings: <4 recommended

    • Rule of 5 Violations: 0-1 preferred for oral drugs

    • QED: Drug-likeness score (0-1, higher is better)

  2. Explain property implications for drug development


Usage Examples#

Example 1: Researching a Drug Target#

You: “I’m interested in EGFR inhibitors. What are the IC50 values of gefitinib and erlotinib against EGFR?”

Claude will:

  1. Use Target Search to find EGFR target ID

  2. Use Compound Search to find gefitinib and erlotinib

  3. Use Get Bioactivity for each compound against EGFR

  4. Compare IC50 values

  5. Explain relative potencies

Example 2: Drug Repurposing#

You: “Find approved drugs that target BCR-ABL. I want to explore repurposing opportunities.”

Claude will:

  1. Use Target Search to find BCR-ABL

  2. Use Drug Search filtered by approved status

  3. Use Get Bioactivity to show potencies

  4. Use Get Mechanism to understand action types

  5. List current indications for each drug

Example 3: Lead Optimization#

You: “I have a compound with ChEMBL ID CHEMBL123456. What are its drug-like properties and how potent is it?”

Claude will:

  1. Use Compound Search to get compound details

  2. Use Get ADMET to show molecular properties

  3. Use Get Bioactivity to find target activities

  4. Assess Rule of 5 violations

  5. Suggest property improvements if needed


Need Help?#

For issues or questions about the ChEMBL Connector, see our Troubleshooting Guide.