# Bio Research

> Anthropic’s life sciences R&D plugin: search papers, compounds, targets and trials, and run single-cell QC and nf-core pipelines.

Anthropic's bio-research plugin, one of its knowledge-work plugins, for early-stage life sciences R&D. It's built for Cowork, Anthropic's desktop agent, and also works in Claude Code. It bundles connectors to literature, compound, drug-target and clinical-trial databases, and skills for single-cell analysis, sequencing pipelines, lab data conversion and choosing which research problem to work on.

What you can do with it
- Search the literature in PubMed, bioRxiv and medRxiv preprints, and Consensus
- Look up bioactive compounds in ChEMBL, find and prioritise drug targets with Open Targets, and review studies on ClinicalTrials.gov
- Run quality control on single-cell RNA-seq data (.h5ad, or 10x Genomics .h5) following scverse best practices, with MAD-based filtering and plots
- Integrate and batch-correct single-cell data, transfer cell-type labels and analyse multi-modal data with scvi-tools models such as scVI, scANVI, totalVI, PeakVI and MultiVI
- Run nf-core pipelines (rnaseq, sarek for variant calling, atacseq) on your own FASTQ files or public GEO/SRA data, from the samplesheet to checking the outputs
- Convert instrument exports (PDF, CSV, Excel, text) from cell counters, plate readers, spectrophotometers, qPCR and chromatography systems into Allotrope Simple Model JSON or a flat CSV, with parser code to hand to data engineers
- Pitch a new project idea, troubleshoot a stuck one or think through a strategic decision with a problem-selection framework based on Fischbach and Walsh's paper in Cell
- Make scientific figures in BioRender

Connectors it bundles
- No account needed: PubMed, bioRxiv (with medRxiv), ClinicalTrials.gov, ChEMBL and Open Targets
- Sign in with your own account: Consensus, BioRender, Synapse (Sage Bionetworks), Wiley Scholar Gateway and Owkin
- Benchling is listed too, but its address is still a placeholder, so it doesn't connect yet
- Optional extras you download yourself: 10x Genomics' txg-mcp and Harvard MIMS's ToolUniverse
Its skills are written for categories rather than products, so another server of the same kind works in place of a bundled one.

How to install it
In Claude Code, run claude plugin marketplace add anthropics/knowledge-work-plugins, then claude plugin install bio-research@knowledge-work-plugins. Its start skill checks which connectors are working and lists the analysis skills.

Good to know
- The nf-core pipelines need Docker, Nextflow and Java on your machine; the skill checks your environment and runs a test profile before the real run.
- The single-cell, scvi-tools and Allotrope skills run bundled Python scripts, so they need Python with packages such as scanpy, scvi-tools and allotropy.
- Anthropic presents the Allotrope converter as an example skill, to adapt to your own schemas.
- The skills are Apache 2.0. Each connector is run by its own provider, under its own terms.

- Type: Plugin
- Made by: Anthropic
- Category: Research
- Formats: Claude plugins
- Version: latest
- Website: https://github.com/anthropics/knowledge-work-plugins/tree/main/bio-research
- Source: https://github.com/anthropics/knowledge-work-plugins/tree/HEAD/bio-research
- Page: https://pluginboard.ai/plugins/anthropic-bio-research

## Works with

- Claude
- GitHub Copilot
- Claude Code
- OpenClaw

## Install

### Claude

In Claude (web or desktop): Customize → Plugins → Add → Add marketplace, enter anthropics/knowledge-work-plugins, then add bio-research

### Claude Code

In Claude Code:

```text
/plugin marketplace add anthropics/knowledge-work-plugins
/plugin install bio-research@knowledge-work-plugins
```

---
From Pluginboard (https://pluginboard.ai).