molecular_targets

Name: molecular_targets
Author: QSong-github

132stars4forks

How to install molecular_targets?

qsong-github/drugclaw/molecular_targets

$npx -y skills add QSong-github/DrugClaw --skill molecular_targets

Installs into the current project.

Summary

Query the NCI CCDI Molecular Targets Platform (pediatric oncology) for targets (genes), diseases, drugs, and target-disease associations via its public GraphQL API. Auto-detects entity type from input string.

Files · 1

View on GitHub

SKILL.md

1	# Molecular Targets Platform Query Skill
2
3	Search the CCDI Molecular Targets Platform — an NCI-supported Open Targets
4	instance focused on preclinical pediatric oncology data. Includes FDA
5	Pediatric Molecular Target Lists and additional pediatric cancer datasets.
6
7	## Entity Auto-Detection
8
9	\| Input Pattern \| Detected As \| Query Used \|
10	\|---\|---\|---\|
11	\| `ENSG00000141510` \| Target (Ensembl gene) \| `target(ensemblId)` \|
12	\| `EFO_\d+`, `MONDO_\d+`, `Orphanet_\d+`, `HP_\d+`, `DOID_\d+` \| Disease / Phenotype \| `disease(efoId)` \|
13	\| `CHEMBL\d+` \| Drug / molecule \| `drug(chemblId)` \|
14	\| anything else \| Free text \| `search(queryString)` \|
15
16	## API
17
18	\| Function \| Input \| Returns \|
19	\|---\|---\|---\|
20	\| `search(entity, size=10)` \| single entity string \| `dict` (GraphQL data) or `None` \|
21	\| `search_batch(entities)` \| list of strings \| `dict[str, dict\\|None]` — uses batch GraphQL where possible \|
22	\| `get_associations(entity, size=10)` \| Ensembl ID or EFO/MONDO ID \| associated diseases (for target) or targets (for disease) \|
23	\| `summarize(data, entity)` \| result dict + label \| compact LLM-readable text \|
24	\| `to_json(data)` \| result dict \| `list[dict]` for pipeline output \|
25
26	### Helper
27
28	\| Function \| Purpose \|
29	\|---\|---\|
30	\| `detect_entity_type(entity)` \| returns `'target'`, `'disease'`, `'drug'`, or `'search'` \|
31
32	## Usage
33
34	See `if __name__ == "__main__"` block in `16_MolecularTargets.py` for runnable
35	examples covering:
36
37	1. Free-text search (`"neuroblastoma"`)
38	2. Target lookup by Ensembl ID (`"ENSG00000141510"` → TP53)
39	3. Disease lookup by ontology ID (`"MONDO_0005072"`)
40	4. Drug lookup by ChEMBL ID (`"CHEMBL941"` → imatinib)
41	5. Target → disease associations
42	6. Disease → target associations
43	7. Batch search (mixed entity types)
44	8. JSON pipeline output
45
46	## Data Source
47
48	- Platform: NCI CCDI Molecular Targets Platform
49	- URL: <https://moleculartargets.ccdi.cancer.gov>
50	- API endpoint: `https://moleculartargets.ccdi.cancer.gov/api/v4/graphql`
51	- Method: HTTP POST, `Content-Type: application/json`, no auth required
52	- Upstream schema: Open Targets Platform GraphQL v4
53	- Focus: Pediatric oncology — FDA Pediatric Molecular Target Lists, preclinical data
54
55	## Notes
56
57	- The API is optimised for single-entity queries. For bulk analyses consider
58	Open Targets data downloads.
59	- Batch functions (`search_batch`) use the plural GraphQL fields
60	(`targets`, `diseases`, `drugs`) to reduce round-trips where possible;
61	free-text queries fall back to one-at-a-time.
62	- `get_associations()` returns scored target↔disease links ranked by overall
63	evidence score (0–1). Use `size` to control how many rows are returned.
64	- Returns `None` / `[]` on HTTP errors — callers handle no-results gracefully.

People who liked this also viewed

6 picks

Type

molecular_targets

132stars4forks

View on GitHub

How to install molecular_targets?

qsong-github/drugclaw/molecular_targets

$npx -y skills add QSong-github/DrugClaw --skill molecular_targets

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Molecular Targets Platform Query Skill
2
3	Search the CCDI Molecular Targets Platform — an NCI-supported Open Targets
4	instance focused on preclinical pediatric oncology data. Includes FDA
5	Pediatric Molecular Target Lists and additional pediatric cancer datasets.
6
7	## Entity Auto-Detection
8
9	\| Input Pattern \| Detected As \| Query Used \|
10	\|---\|---\|---\|
11	\| `ENSG00000141510` \| Target (Ensembl gene) \| `target(ensemblId)` \|
12	\| `EFO_\d+`, `MONDO_\d+`, `Orphanet_\d+`, `HP_\d+`, `DOID_\d+` \| Disease / Phenotype \| `disease(efoId)` \|
13	\| `CHEMBL\d+` \| Drug / molecule \| `drug(chemblId)` \|
14	\| anything else \| Free text \| `search(queryString)` \|
15
16	## API
17
18	\| Function \| Input \| Returns \|
19	\|---\|---\|---\|
20	\| `search(entity, size=10)` \| single entity string \| `dict` (GraphQL data) or `None` \|
21	\| `search_batch(entities)` \| list of strings \| `dict[str, dict\\|None]` — uses batch GraphQL where possible \|
22	\| `get_associations(entity, size=10)` \| Ensembl ID or EFO/MONDO ID \| associated diseases (for target) or targets (for disease) \|
23	\| `summarize(data, entity)` \| result dict + label \| compact LLM-readable text \|
24	\| `to_json(data)` \| result dict \| `list[dict]` for pipeline output \|
25
26	### Helper
27
28	\| Function \| Purpose \|
29	\|---\|---\|
30	\| `detect_entity_type(entity)` \| returns `'target'`, `'disease'`, `'drug'`, or `'search'` \|
31
32	## Usage
33
34	See `if __name__ == "__main__"` block in `16_MolecularTargets.py` for runnable
35	examples covering:
36
37	1. Free-text search (`"neuroblastoma"`)
38	2. Target lookup by Ensembl ID (`"ENSG00000141510"` → TP53)
39	3. Disease lookup by ontology ID (`"MONDO_0005072"`)
40	4. Drug lookup by ChEMBL ID (`"CHEMBL941"` → imatinib)
41	5. Target → disease associations
42	6. Disease → target associations
43	7. Batch search (mixed entity types)
44	8. JSON pipeline output
45
46	## Data Source
47
48	- Platform: NCI CCDI Molecular Targets Platform
49	- URL: <https://moleculartargets.ccdi.cancer.gov>
50	- API endpoint: `https://moleculartargets.ccdi.cancer.gov/api/v4/graphql`
51	- Method: HTTP POST, `Content-Type: application/json`, no auth required
52	- Upstream schema: Open Targets Platform GraphQL v4
53	- Focus: Pediatric oncology — FDA Pediatric Molecular Target Lists, preclinical data
54
55	## Notes
56
57	- The API is optimised for single-entity queries. For bulk analyses consider
58	Open Targets data downloads.
59	- Batch functions (`search_batch`) use the plural GraphQL fields
60	(`targets`, `diseases`, `drugs`) to reduce round-trips where possible;
61	free-text queries fall back to one-at-a-time.
62	- `get_associations()` returns scored target↔disease links ranked by overall
63	evidence score (0–1). Use `size` to control how many rows are returned.
64	- Returns `None` / `[]` on HTTP errors — callers handle no-results gracefully.

People who liked this also viewed

6 picks

Type