Search PubMed for scientific literature. Supports full NCBI query syntax (MeSH terms, Boolean operators, field tags, date ranges). Returns articles with title, authors, abstract, DOI, PMID, journal, year, and MeSH terms. Results auto-indexed into session knowledge graph.

Retrieve gene information from NCBI Gene — symbol, full name, chromosomal location, aliases, RefSeq IDs, and functional summary. Auto-indexes gene entity into session knowledge graph.

Run NCBI BLAST sequence alignment (blastp/blastn/blastx/tblastn). Async polling — waits up to 120s for results.

Full UniProt Swiss-Prot entry: function, domains, PTMs, GO terms, disease links, sequence. Prefer reviewed accessions (P/Q/O prefix). Auto-indexes protein + disease edges into session knowledge graph.

Unified protein discovery across UniProt and PDB. Use accession for a direct record lookup or query to search reviewed proteins and experimental structures.

AlphaFold DB predicted structure: per-residue pLDDT confidence stats, PDB/mmCIF download URLs. pLDDT ≥90=very high, 70–90=confident, <50=disordered.

Merged pathway workflow. Search KEGG, retrieve genes for a pathway, or assemble Reactome plus KEGG context for a gene.

ChEMBL drug-target activities: IC50, Ki, Kd values, assay types, approval status. Auto-indexes drug→gene edges into knowledge graph.

Open Targets gene-disease evidence across 6 datatypes: genetic_association, somatic_mutation, known_drug, animal_model, affected_pathway, literature. Auto-indexes gene→disease edges into knowledge graph.

ClinicalTrials.gov v2: trial status, phase, interventions, enrollment, eligibility. Auto-indexes drug→disease treatment edges from trials.

FLAGSHIP: 7-database parallel integration — NCBI Gene, PubMed, Reactome, ChEMBL, Open Targets, GEO, ClinicalTrials.gov. One call, complete overview.

Boltz-2 structure workflow. Use mode='structure' for direct multimolecular structure prediction or mode='protein_ligand' for the integrated UniProt-to-docking workflow.

Evo2 DNA workflow. Use mode='generate' for sequence generation or mode='score' for wildtype-versus-variant scoring.

Merged CRISPR workflow covering guide design, guide scoring, off-target review, base editing, and repair outcome estimation.

Merged FDA drug-safety workflow for adverse-event search, signal detection, label review, and head-to-head comparison.

Merged variant-interpretation workflow for ACMG classification, population frequency, ClinVar review, splice review, and full integrated reporting.

Drug repurposing engine: surface approved drugs with activity against a target/disease.

Verify a biological claim with structured claim decomposition and relation-specific database evidence.

cBioPortal cancer mutation frequencies across TCGA cohorts.

NHGRI-EBI GWAS Catalog: genome-wide significant associations for a gene.

Analyze one or two gene panels in parallel and return a cross-gene comparison matrix. In differential mode it ranks pathways and diseases enriched in panel A versus panel B using panel-level hit fractions and fold-change-style scoring.

Merged research-session workflow for entity resolution, live graph inspection, persisted graph save/restore via MCP resources, provenance export, and automated planning.

Check FDA label interaction context between a primary drug and a list of co-medications.

Summarize candidate binding sites from UniProt functional-site annotations plus AlphaFold confidence context.

Draft a disease-focused biomarker panel using Open Targets evidence with a literature fallback.

Summarize CPIC-style pharmacogenomic genes and supporting PGx evidence for a drug.

Summarize protein family and domain context from curated UniProt annotations with direct Pfam and InterPro links.

Summarize recurrent Reactome pathways and STRING network hubs across a gene set.

Marker-based heuristic deconvolution of a bulk RNA-seq profile into likely cell-type fractions.

PubChem-backed structural similarity search from a compound name or SMILES string.

Normalize phenotype terms and rank OMIM differentials for a candidate gene.

Generate genome-browser links and locus context for a gene or explicit genomic interval.