Analyze RNAseq Data End-to-End with Claude Code

A 1-hour, hands-on workshop. You bring a terminal and curiosity. By the end you’ll have taken a real public bulk-RNAseq dataset from salmon quantifications to a finished differential-expression analysis — with heatmaps, volcano plots, and GO enrichment — driven entirely by prompting Claude Code.

What you’ll learn

  • How to point Claude Code at a real bioinformatics dataset and have it understand the data before writing any code.
  • The standard bulk-RNAseq downstream workflow — tximport → DESeq2 → results → visualization → enrichment — and how to drive each step in plain English.
  • How to review, correct, and trust AI-generated analysis code instead of copy-pasting blindly.
  • How preprocessing (FASTQ → salmon quant.sf) works, so you can reproduce the whole pipeline later.

The dataset

A real, published experiment from GEO: GSE197576 — human cells under hypoxia vs normoxia. Four single-end samples (2 per condition). We expect classic hypoxia-induced genes (VEGFA, CA9, BNIP3…) to light up — a built-in sanity check that the analysis is correct.

Agenda (60 minutes)

Time Segment Page
0–5 Intro to Claude Code, clone the repo Setup
5–12 Explore the salmon data with Claude Code 1 · The data
12–20 How the quant.sf files were made 2 · Preprocessing
20–42 Import & differential expression 3 · Import & DESeq2
42–54 Visualize the results 4 · Visualization
54–60 Functional enrichment + wrap-up 5 · Enrichment · 6 · Wrap-up

How this site works

Every analysis step shows two things: the natural-language prompt you give Claude Code, and the code/result it produces. The prompts look like this:

Try this prompt

Load the salmon quantification files in data/salmon/ with tximport, summarize them to gene level using data/tx2gene.csv, and build a DESeq2 dataset comparing hypoxia vs normoxia.

You don’t have to type them verbatim — they’re a starting point. Claude Code reads CLAUDE.md in this repo, so it already knows where the data lives and what we’re trying to do.

Start at Setup.

Back to top