Claude Code Biology Research

🧬 Claude Code for Biology Research

A comprehensive guide to using Claude Code as an AI-powered assistant for computational biology, bioinformatics, and biomedical research

---

📋 Table of Contents

• [🌐 Overview](#-overview)
• [✨ Key Features](#-key-features)
• [📦 Installation](#-installation)
• [System Requirements](#system-requirements)
• [Installing Claude Code](#installing-claude-code)
• [Authentication](#authentication)
• [Proxy Configuration](#proxy-configuration)
• [Verifying Installation](#verifying-installation)
• [🚀 Quick Start](#-quick-start)
• [Your First Biology Session](#your-first-biology-session)
• [Key CLI Commands](#key-cli-commands)
• [🧠 Skills for Biology Research](#-skills-for-biology-research)
• [Built-in Skills](#built-in-skills)
• [Custom Biology Skills](#custom-biology-skills)
• [Skill Development Guide](#skill-development-guide)
• [🔌 MCP Servers](#-mcp-servers)
• [What is MCP?](#what-is-mcp)
• [Essential Biology MCP Servers](#essential-biology-mcp-servers)
• [MCP Configuration Guide](#mcp-configuration-guide)
• [Custom MCP Server Development](#custom-mcp-server-development)
• [🧪 Biology Workflows](#-biology-workflows)
• [Genomics & Transcriptomics](#genomics--transcriptomics)
• [Proteomics & Structural Biology](#proteomics--structural-biology)
• [Literature Review & Meta-analysis](#literature-review--meta-analysis)
• [Biological Image Analysis](#biological-image-analysis)
• [Laboratory Protocol Design](#laboratory-protocol-design)
• [📝 Practical Examples](#-practical-examples)
• [Example 1: RNA-seq Data Analysis](#example-1-rna-seq-data-analysis)
• [Example 2: Protein Structure Retrieval](#example-2-protein-structure-retrieval)
• [Example 3: Literature Review Automation](#example-3-literature-review-automation)
• [⚙️ Advanced Configuration](#️-advanced-configuration)
• [Settings.json Deep Dive](#settingsjson-deep-dive)
• [Custom Hooks](#custom-hooks)
• [Environment Management](#environment-management)
• [🤝 Best Practices](#-best-practices)
• [📚 Resources & References](#-resources--references)
• [📄 License](#-license)

---

🌐 Overview

Claude Code is Anthropic's official command-line interface (CLI) for Claude, designed to be an agentic AI coding assistant that works directly in your terminal. For biology researchers and computational biologists, Claude Code represents a paradigm shift — it bridges the gap between biological domain knowledge and computational implementation.

Instead of switching between multiple tools, databases, and programming environments, you can:

1. Describe your biology research task in natural language
2. Let Claude Code understand the context and generate appropriate code
3. Execute, debug, and refine analyses interactively
4. Automate repetitive bioinformatics workflows
5. Integrate with public biological databases and lab data systems

How It Works

Claude Code operates as an agentic system in your terminal:

You: "Download RNA-seq data from GEO accession GSE12345, perform differential expression analysis, and create a volcano plot"

Claude Code:
  ✓ Parses your request into actionable steps
  ✓ Uses MCP tools to query GEO database
  ✓ Downloads FASTQ/count data
  ✓ Writes Python/R script for DESeq2 analysis
  ✓ Executes the analysis
  ✓ Generates publication-quality figures
  ✓ Summarizes results

This guide covers everything from installation to advanced workflow orchestration for biology research.

---

✨ Key Features

| Feature | Biology Research Application | |---------|---------------------------| | 🧬 Natural Language to Code | Describe biology concepts; Claude writes Python/R/Bash scripts | | 🔗 MCP Tool Integration | Direct access to NCBI, UniProt, PDB, PubMed, and more | | 📊 Multi-modal Support | Analyze microscope images, gel photos, and plots | | 🔄 Autonomous Workflows | End-to-end analysis pipelines with iterative refinement | | 📝 Academic Writing | Draft manuscripts, methods sections, and literature reviews | | 📦 Reproducible Science | Generate Conda/Docker environments, package requirements | | 🗂️ Project Management | Organize code, data, and results in structured projects | | 🤝 Collaborative | Git-native, team-friendly, shareable analysis scripts |

---

📦 Installation

System Requirements

| Requirement | Minimum | Recommended | |------------|---------|-------------| | OS | macOS 10.15+, Ubuntu 20.04+, Windows 10+ (via WSL2) | Latest stable | | Node.js | 18.x | 20.x LTS | | npm | 8.x | 10.x+ | | Python | 3.8 | 3.11+ (for bioinformatics) | | R | 4.0 | 4.3+ (for Bioconductor) | | Disk Space | 1 GB | 10 GB+ (for biological databases) | | RAM | 4 GB | 16 GB+ | | Internet | Required for installation and API access | Broadband |

Installing Claude Code

Method 1: npm (Recommended)

# Install globally
npm install -g @anthropic-ai/claude-code

# Or use npx for one-off usage
npx @anthropic-ai/claude-code

Method 2: VS Code Extension

Install the Claude Code Extension from the VS Code Marketplace, or:

# Install via VS Code CLI
code --install-extension anthropic.claude-code

Method 3: Homebrew (macOS/Linux)

brew install anthropic/claude-code/claude-code

Method 4: Direct Download

Download the latest release from Claude Code Releases.

Authentication

Claude Code requires an Anthropic API key:

# Set your API key
export ANTHROPIC_API_KEY="sk-ant-xxxxxxxxxxxx"

# Or save it to ~/.claude/.env for persistence
echo 'ANTHROPIC_API_KEY="sk-ant-xxxxxxxxxxxx"' >> ~/.claude/.env

> 💡 For biology researchers: If your institution has an enterprise Anthropic account, request an API key through your institutional IT department. Consider security and data privacy — do not share your API key.

Getting Your API Key

1. Visit console.anthropic.com
2. Create an account or sign in
3. Navigate to API Keys
4. Create a new key
5. Copy the key and store it securely

Proxy Configuration

If your institution uses a proxy (common in research environments):

# HTTP/HTTPS proxy
export HTTP_PROXY="http://proxy.your-institution.edu:8080"
export HTTPS_PROXY="http://proxy.your-institution.edu:8080"

# Configure git proxy (for cloning repos)
git config --global http.proxy http://proxy.your-institution.edu:8080
git config --global https.proxy http://proxy.your-institution.edu:8080

# Configure npm proxy
npm config set proxy http://proxy.your-institution.edu:8080
npm config set https-proxy http://proxy.your-institution.edu:8080

Verifying Installation

# Check version
claude --version

# Start an interactive session
claude

# Run a quick test
claude -p "Write a Python script that prints the reverse complement of a DNA sequence: ATGCGTAC"

Expected output:

def reverse_complement(seq):
    """Return the reverse complement of a DNA sequence."""
    complement = {'A': 'T', 'T': 'A', 'C': 'G', 'G': 'C'}
    return ''.join(complement.get(base, base) for base in reversed(seq))

seq = "ATGCGTAC"
result = reverse_complement(seq)
print(f"Original: {seq}")
print(f"Reverse complement: {result}")
# Output: Original: ATGCGTAC
#         Reverse complement: GTACGCAT

---

🚀 Quick Start

Your First Biology Session

# Start Claude Code
claude

Once inside the interactive session, try these biology research prompts:

1️⃣ "Download the human genome assembly hg38 FASTA file using curl and index it with samtools"

2️⃣ "Write a Python script using BioPython to parse a GenBank file and extract CDS features.
     Save it as parse_genbank.py"

3️⃣ "Search PubMed for recent papers on CRISPR prime editing and summarize the top 5"

4️⃣ "Create a volcano plot from this differential expression CSV file using ggplot2 in R"

Key CLI Commands

| Command | Purpose | Biology Research Use | |---------|---------|---------------------| | claude | Start interactive session | Begin a research analysis session | | claude -p "prompt" | Run one-off command | Quick sequence manipulation | | claude --model "claude-sonnet-4-20250514" | Specify model | Use latest model for complex analyses | | claude -e "python script.py" " | Edit existing file | Refine an analysis script | | claude -i "path/to/file" | Include system prompt | Load institutional protocols | | /clear | Clear session | Reset for new analysis | | /help | Show help | Browse available commands | | /init | Init project config | Set up project-specific settings | | /fast | Toggle fast mode | Quick responses for simple tasks | | /loop | Repeat command periodically | Monitor long-running analyses |

> 💡 Pro tip: Use claude -p "prompt" in shell scripts to build automated bioinformatics pipelines!

---

🧠 Skills for Biology Research

Claude Code's Skills are specialized capabilities — think of them as plugins that give Claude expert-level abilities in specific domains. Skills can be loaded via /skill-name in CLI or invoked programmatically.

Built-in Skills

Here are the most valuable built-in skills for biology research:

📚 Literature & Writing Skills

| Skill | Invocation | Description | Biology Use Case | |-------|-----------|-------------|-----------------| | systematic-literature-review | /systematic-literature-review | Multi-source literature search, quality scoring, and auto-generation of structured reviews | Systematic reviews, meta-analyses, related work sections | | academic-writing-polisher | /academic-writing-polisher | Polish academic writing with domain-aware style corrections | Manuscript refinement, grant proposals | | paper-write-sci | /paper-write-sci | Generate scientific paper drafts from data and outlines | Full manuscript drafting | | paper-select-journal | /paper-select-journal | Recommend target journals based on manuscript content | Journal selection guidance | | review-papers | /review-papers | Peer review simulation with structured feedback | Pre-submission review | | paper-explain-figures | /paper-explain-figures | Interpret scientific figures and create accessible descriptions | Figure legends, accessibility |

🔬 Bioinformatics & Lab Skills

| Skill | Invocation | Description | |-------|-----------|-------------| | chem-vis | /chem-vis | Generate 2D/3D molecular visualizations from SMILES/names | | thesis-review | /thesis-review | Comprehensive thesis structure and content review | | systematic-literature-review | /systematic-literature-review | AI-powered literature survey with quality scoring |

🛠️ Development & Workflow Skills

| Skill | Invocation | Description | |-------|-----------|-------------| | code-review | /code-review | Review bioinformatics code for correctness and efficiency | | simplify | /simplify | Optimize and refactor analysis scripts | | verify | /verify | Verify that code changes work as expected | | deep-research | /deep-research | Multi-source deep research with fact-checking and citations | | init | /init | Project scaffolding | | run | /run | Launch and interact with project applications |

Custom Biology Skills

You can create custom skills for your specific biology research needs:

# Create a new skill
claude -p "Create a custom skill for single-cell RNA-seq analysis workflow automation"

Example custom skill structure for a single-cell RNA-seq analysis skill:

📁 Custom Skill Example: scRNA-seq Analysis Skill (Click to expand)

# Skill: scrna-seq-analysis

## Description
Automated single-cell RNA-seq analysis pipeline using ScanPy and Seurat

## Triggers
- "analyze single cell data"
- "scRNA-seq workflow"
- "single cell clustering"

## Steps
1. Quality control: Filter cells based on gene counts and mitochondrial content
2. Normalization: Log-normalize and identify highly variable genes
3. Dimensionality reduction: PCA, UMAP, t-SNE
4. Clustering: Leiden clustering with resolution optimization
5. Marker identification: Differential expression per cluster
6. Cell type annotation: Automated annotation using reference databases
7. Visualization: UMAP plots, heatmaps, dotplots

## Dependencies
- scanpy>=1.9
- anndata>=0.9
- leidenalg
- matplotlib>=3.6
- pandas>=1.5

Skill Development Guide

To create biology-focused skills:

# Step 1: Use the skill-creator
# In Claude Code:
/skill-creator

# Step 2: Describe your biology workflow
# "Create a skill for ChIP-seq peak calling and motif analysis"

# Step 3: The skill will be registered and available via /your-skill-name

Skill Design Principles for Biology:

| Principle | Description | Example | |-----------|-------------|---------| | Domain-specific | Focus on one biological analysis type | "ChIP-seq analysis" not "sequencing analysis" | | Reproducible | Include version-locked dependencies | scanpy==1.9.3 | | Documented | Clear input/output specifications | Input: FASTQ files, Output: peak BED files | | Modular | Composable with other skills | QC skill → Alignment skill → Peak calling skill |

---

🔌 MCP Servers

What is MCP?

MCP (Model Context Protocol) is an open standard that allows Claude Code to connect directly with external tools, databases, and services. Think of MCP servers as "adapters" that give Claude real-time access to biological data sources and computational tools.

How MCP works with Claude Code:

┌─────────────┐    MCP Protocol    ┌──────────────┐    HTTP/API    ┌──────────────┐
│  Claude Code │ ◄──────────────► │  MCP Server  │ ◄────────────► │  NCBI / UniProt / PDB / PubMed  │
│  (AI Engine) │                  │  (Adapter)   │               │  (Databases)  │
└─────────────┘                  └──────────────┘               └──────────────┘

Essential Biology MCP Servers

1️⃣ NCBI/BLAST MCP Server

Access National Center for Biotechnology Information databases and BLAST search:

{
  "mcpServers": {
    "ncbi-biology": {
      "command": "uvx",
      "args": ["ncbi-mcp-server"],
      "env": {
        "NCBI_API_KEY": "your_ncbi_api_key",
        "EMAIL": "your.email@institution.edu"
      }
    }
  }
}

Capabilities:

• search_nucleotide(query) — Search nucleotide databases
• search_protein(query) — Search protein databases
• run_blast(sequence, database) — BLAST sequence search
• fetch_genbank(accession) — Download GenBank records
• fetch_fasta(accession) — Download FASTA sequences
• search_pubmed(query) — PubMed literature search
• search_geo(query) — GEO dataset search

> 🔑 NCBI API Key: Register at ncbi.nlm.nih.gov for higher rate limits. Without a key, you're limited to 3 requests/second; with a key, 10 requests/second.

2️⃣ UniProt/PDB MCP Server

Access protein sequence, structure, and functional data:

{
  "mcpServers": {
    "uniprot-pdb": {
      "command": "uvx",
      "args": ["uniprot-mcp-server"]
    }
  }
}

Capabilities:

• search_uniprot(query) — Search UniProtKB
• get_protein(accession) — Retrieve protein entry
• fetch_pdb(pdb_id) — Download PDB structure files
• search_pdb(query) — Search PDB for structures
• get_alphafold(uniprot_id) — Retrieve AlphaFold predictions
• get_protein_features(accession) — Domain, site, and feature annotations
• search_by_sequence(sequence) — Sequence similarity search

3️⃣ PubMed MCP Server

Advanced literature mining and retrieval:

{
  "mcpServers": {
    "pubmed-mcp": {
      "command": "npx",
      "a

…

## Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [Jeny-Liu](https://github.com/Jeny-Liu)
- **Source:** [Jeny-Liu/claude-code-biology-research](https://github.com/Jeny-Liu/claude-code-biology-research)
- **License:** MIT
- **Homepage:** https://github.com/Jeny-Liu/claude-code-biology-research

Install and usage instructions live in the source repository linked above.