Skip to main content

​
Overview

The Bio-Research plugin consolidates 10 MCP server integrations and 5 analysis skills into a single package for life science researchers. Search biomedical literature, run genomics pipelines, analyze single-cell data, and standardize instrument outputs — all in one place. Designed for Cowork and Claude Code.

​
Installation

/install anthropics/knowledge-work-plugins bio-research

​
What’s Included

​
MCP Servers (Data Sources & Tools)

The plugin connects to 10 MCP servers across literature, drug discovery, and scientific tools:

Literature & Data

  • PubMed — Biomedical literature search
  • bioRxiv/medRxiv — Preprint access
  • Wiley Scholar Gateway — Academic publications
  • Synapse — Collaborative research data management

Drug Discovery

  • ChEMBL — Bioactive compound database
  • Open Targets — Drug target discovery
  • ClinicalTrials.gov — Clinical trial registry

Visualization & AI

  • BioRender — Scientific illustration creation
  • Owkin — AI for histopathology and drug discovery

Lab Platform

  • Benchling — Lab data management (placeholder)
Plugin files use ~~category placeholders (e.g., ~~literature, ~~chemical database) that resolve to whichever MCP server you connect in that category. See CONNECTORS.md for alternative options.

​
Optional Binary MCP Servers

These require separate binary downloads:
  • 10X Genomics txg-mcp (~~genomics platform) — Cloud analysis data and workflows
    Download from GitHub
  • ToolUniverse (~~tool database) — AI tools for scientific discovery from Harvard MIMS
    Download from GitHub

​
Analysis Skills

Five specialized skills power your research workflows:
Automated quality control for scRNA-seq data following scverse best practices.Features:
  • Supports .h5ad and .h5 file formats
  • MAD-based filtering for cells and genes
  • Comprehensive visualizations (violin plots, scatter plots, highest expressed genes)
  • Mitochondrial content analysis
  • Doublet detection
Output: Filtered dataset ready for downstream analysis with QC report
Deep learning toolkit for single-cell omics analysis.Supported models:
  • scVI — Dimensionality reduction and batch correction
  • scANVI — Semi-supervised cell type annotation with label transfer
  • totalVI — Joint RNA and protein analysis (CITE-seq)
  • PeakVI — scATAC-seq analysis
  • MultiVI — Multi-modal integration
  • DestVI — Deconvolution of spatial transcriptomics
  • veloVI — RNA velocity with uncertainty quantification
  • sysVI — Systems biology modeling
Use cases: Integration, batch correction, label transfer, multi-modal analysis
Run nf-core bioinformatics pipelines on local or public sequencing data.Available pipelines:
  • rnaseq — Gene expression quantification and differential expression analysis
  • sarek — Germline and somatic variant calling (WGS/WES)
  • atacseq — Chromatin accessibility analysis
Data sources: Local FASTQ files or public GEO/SRA datasetsAutomatically handles pipeline configuration, execution, and output verification.
Convert laboratory instrument output files to Allotrope Simple Model (ASM) format.Supported formats: PDF, CSV, Excel, TXTInstrument types (40+):
  • Cell counters and viability analyzers
  • Spectrophotometers (UV-Vis, fluorescence)
  • Plate readers
  • qPCR and digital PCR systems
  • Chromatography systems (HPLC, LC-MS)
  • Flow cytometers
  • Thermal cyclers
Enables standardized data interchange and FAIR data principles.
Systematic framework for research problem selection based on Fischbach & Walsh’s methodology.9 sub-skills covering:
  • Ideation and brainstorming
  • Risk assessment and feasibility analysis
  • Resource optimization
  • Decision trees for prioritization
  • Adversity planning (what could go wrong)
  • Synthesis and final selection
Use when pitching a new idea, troubleshooting a stuck project, or evaluating strategic decisions.

​
Commands

​
/start

Initialize your bio-research environment and explore available tools. 
/bio-research:start
What it does:
  1. Displays a welcome message with plugin overview
  2. Tests which MCP servers are connected and reports their status
  3. Lists available analysis skills with descriptions
  4. Mentions optional binary MCP servers (10X Genomics, ToolUniverse)
  5. Asks what you’re working on and suggests starting points
Common starting points:
  • Literature review
  • Analyze sequencing data (QC, pipelines)
  • Drug discovery (compound search, target prioritization)
  • Data standardization (Allotrope conversion)
  • Research strategy evaluation

​
Common Workflows

Literature Review

Search ~~literature for "CRISPR gene editing"
Access full text via ~~journal access
Create figure with ~~scientific illustration
Seamlessly search PubMed, access papers through Wiley, and generate publication-ready figures with BioRender.

Single-Cell Analysis

Run QC on scRNA-seq data in my_data.h5ad
Use scvi-tools for batch correction
Perform cell type annotation with scANVI
Complete pipeline from raw counts to annotated cell types with deep learning.

Sequencing Pipeline

Download GEO dataset GSE12345
Run nf-core/rnaseq pipeline
Generate differential expression report
Automated pipeline execution from public or local data with validated results.

Drug Discovery

Search ~~chemical database for CDK inhibitors
Use ~~drug targets for target prioritization
Review clinical trial data for safety profile
Integrated workflow from compound discovery to clinical validation.

​
Example: Complete RNA-seq Analysis

You: I have RNA-seq data from a GEO dataset. Can you help me analyze it?

Claude: I'll help you run the nf-core/rnaseq pipeline. What's the GEO accession?

You: GSE147507

Claude: [Downloads data, configures pipeline, runs analysis]
        Analysis complete. Found 1,234 differentially expressed genes.
        Results saved to results/differential_expression/

​
Example: Literature Search with Visualization

You: Find recent papers on CAR-T cell therapy and create a summary figure

Claude: [Searches PubMed for recent CAR-T papers]
        Found 45 relevant papers from the last 6 months.
        Top finding: New CD19/CD22 dual-targeting approach shows 85% response rate.
        
        [Creates figure in BioRender showing mechanism of action]
        Figure ready: car-t-mechanism.png

​
Connector Reference

CategoryPlaceholderIncludedAlternatives
Literature~~literaturePubMed, bioRxivGoogle Scholar, Semantic Scholar
Scientific illustration~~scientific illustrationBioRender—
Clinical trials~~clinical trialsClinicalTrials.govEU Clinical Trials Register
Chemical database~~chemical databaseChEMBLPubChem, DrugBank
Drug targets~~drug targetsOpen TargetsUniProt, STRING
Data repository~~data repositorySynapseZenodo, Dryad, Figshare
Journal access~~journal accessWiley Scholar GatewayElsevier, Springer Nature
AI research~~AI researchOwkin—
Lab platform~~lab platformBenchling*—
*Placeholder — MCP URL not yet configured See CONNECTORS.md for details on alternative MCP servers.

​
Getting Help

Run /bio-research:start anytime to:
  • Check which MCP servers are connected
  • See available analysis skills
  • Get suggested workflows for your research

​
License

Skills are licensed under Apache 2.0. MCP servers are provided by their respective authors — see individual server documentation for terms.

Build docs developers (and LLMs) love

Get started for freeTalk to us