tooluniverse-target-research

Provide a gene symbol, UniProt accession, or Ensembl ID and the skill resolves all identifiers, then runs eight research paths covering structure, function, expression, variants, druggability, safety, and literature. The output is a 14-section markdown report with evidence grades and citations.

It queries over 225 tools spanning UniProt, PDB, AlphaFold, ChEMBL, Open Targets, GTEx, Human Protein Atlas, ClinVar, gnomAD, STRING, IntAct, Reactome, KEGG, PubMed, DisGeNET, Pharos, DepMap, BindingDB, and GPCRdb.

It uses documented fallback chains, such as HPA expression when GTEx fails or Open Targets interactions when IntAct returns nothing. Failed tools and empty results are explicitly documented in the report rather than silently skipped.

Yes, every disease claim and key paper is graded on a T1-T4 evidence tier scale, from direct mechanistic evidence down to computational predictions. Each section includes an evidence quality summary and a completeness audit runs before finalizing.

Yes, the skill supports parallel target analysis, running tractability, drugs, diseases, safety, and interaction queries for multiple targets concurrently and producing a comparison table, as demonstrated with EGFR versus HER2.

Results depend on database coverage, so novel or understudied targets may have sparse PPI, expression, or literature data. Some sources like DisGeNET require an API key, and the report includes a data gaps table when minimum requirements are not met.