DrugMechDB Query Skill

SkillDatabases & data

Query the DrugMechDB drug mechanism-of-action database. Use whenever the user asks about drug mechanisms, drug-to-disease paths, biological targets of a drug, or wants to look up any biomedical entity (drug name, protein, disease, DrugBank ID, MESH ID, UniProt ID, GO term, etc.) in DrugMechDB.

Instructions available. Your AI can read the instructions. Execution depends on the setup they require.

Account requirements not reviewed. Check the skill instructions before use; ahel provides instructions and does not run this skill.

Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.

Then ask your AI: use the DrugMechDB Query Skill skill

What this skill tells your AI

The instructions your AI receives, as published by qsong-github/drugclaw in skills/drug_mechanism/drugmechdb/SKILL.md and read by ahel’s review.

Search drug mechanism-of-action paths by entity name or ID. Each path is a directed graph: Drug → (intermediates) → Disease, with typed nodes and labeled edges.

Data

Source: indication_paths.json
Path: resources_metadata/drug_mechanism/DRUGMECHDB/indication_paths.json
Records: ~4846 mechanism paths, ~32k relationships

Entity auto-detection

Input pattern	Detected as	Example
`DB:DB00619`	DrugBank ID	exact on node/graph IDs
`MESH:D015464`	MESH ID	exact on node/graph IDs
`UniProt:P00519`	UniProt protein	exact on node IDs
`GO:0006915`	GO term	exact on node IDs
`CHEBI:`, `HP:`, `UBERON:`, `CL:`, `reactome:`, `InterPro:`, `PR:`, `taxonomy:`	respective types	exact on node IDs
anything else	free text	substring match on drug/disease/node names

API

Function	Signature	Returns
`load(path)`	path to JSON	`list[dict]` — full database
`build_index(db)`	loaded db	`(by_id, by_name, by_drug, by_disease)` dicts for O(1) lookup
`search(db, entity, index=None)`	single query string	`list[dict]` — matching paths
`search_batch(db, entities, index=None)`	list of query strings	`dict[str, list[dict]]`
`summarize(paths, entity)`	search results	compact multi-line text
`to_json(paths)`	search results	list of flat dicts (id, drug, disease, nodes, links)

Node types (14)

BiologicalProcess, Cell, CellularComponent, ChemicalSubstance, Disease, Drug, GeneFamily, GrossAnatomicalStructure, MacromolecularComplex, MolecularActivity, OrganismTaxon, Pathway, PhenotypicFeature, Protein

Quick usage

import drugmechdb_query as dq

db = dq.load()  # uses default DATA_PATH
idx = dq.build_index(db)  # optional, recommended for repeated queries

# Single query — by name or ID
paths = dq.search(db, "imatinib", idx)
paths = dq.search(db, "UniProt:P00519", idx)
paths = dq.search(db, "MESH:D003920", idx)
print(dq.summarize(paths, "imatinib"))

# Batch query
results = dq.search_batch(db, ["metformin", "MESH:D003920", "asthma"], idx)
for entity, paths in results.items():
    print(dq.summarize(paths, entity))

# JSON export
print(dq.to_json(paths))

Output structure per path

graph:  { _id, drug, disease, drugbank, drug_mesh, disease_mesh }
nodes:  [{ id, label, name }, ...]
links:  [{ source, target, key }, ...]

key examples: decreases activity of, causes, positively regulates, treats, increases expression of, etc. (66 relation types total).

Signals

GitHub stars: 116
Forks: 3
Last commit: Aug 2026

Advanced

Item type: skill
Key: drugmechdb-query
Source: github.com/qsong-github/drugclaw

github.com/qsong-github/drugclaw