Guide · Agent skill
How to use the Bulk RNA-seq Preparation agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The bulk-rnaseq skill is an agent skill that prepares bulk RNA-seq data for differential expression. It checks your sample sheet, guides you from raw FASTQ files to a gene-level integer counts matrix using either nf-core/rnaseq or standalone STAR, Salmon and featureCounts recipes, and hands the result to downstream statistics tools. This guide explains what the skill does, how to install it, how to ask for it in plain language, and where its limits are.
Bulk RNA-seq PreparationK-Dense-AI/scientific-agent-skills · MIT Download the skill Quick start
- 1Install the skill. Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 2Prepare your environment. Install Python with pandas and numpy. Add pytximport if you import Salmon output, and set up Nextflow with a container engine (Path A) or bioconda tools (Path B).
- 3Validate the samplesheet. Run the validator with the nf-core option to catch missing FASTQ files, strandedness and replicate problems before any compute is spent.
- 4Choose a quantification path. Use nf-core/rnaseq for audited, scalable runs, or standalone STAR, Salmon or featureCounts for transparency. Use one route for the whole comparison.
- 5Build the counts matrix. For standalone output, run the bridge script to create counts.csv, a metadata template and a provenance file. Fill in the metadata with your design.
- 6Hand off to differential expression. Pass the counts matrix and completed metadata to the pydeseq2 skill for statistical testing, then to pathway-enrichment for pathway analysis.
What an agent skill is and what this one does
An agent skill is a folder of instructions and helper scripts that an AI agent loads when a task matches the skill's description. Once loaded, the agent follows the skill's workflow rather than improvising.
This skill covers the stage between raw sequencing reads and statistical testing. It checks sample sheets for missing or inconsistent FASTQ files, strandedness and replicate problems, then guides the reads-to-counts step. Bundled Python scripts validate the sample sheet and assemble the counts matrix, rejecting ambiguous inputs. It writes a counts file, a metadata template to fill in, and a provenance file with input hashes and import settings. It does not perform the differential expression testing itself; that is handed to the pydeseq2 skill.
- Validates the sample sheet before you spend compute on a run
- Chooses between nf-core/rnaseq (Path A) and standalone tools (Path B)
- Builds a gene x sample integer counts matrix and a metadata template
- Records provenance so the analysis can be reproduced and written up
Install the skill from the ZIP or the Claude apps
Download the ZIP for the skill from this page. You can install it in one of three ways.
For a personal install, unzip it into your user skills folder at ~/.claude/skills/. For a single project, unzip it into the .claude/skills/ folder inside that project. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
The skill's own setup section expects Python with pandas and numpy. The bridge script also needs pytximport, which is only used when importing Salmon output. Path A needs Nextflow with a container engine; Path B needs bioconda tools such as STAR, Salmon, fastp and subread. Network access is required for installation and reference downloads.
- Personal: unzip into ~/.claude/skills/
- Project: unzip into .claude/skills/ inside the project
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Trigger the skill with natural requests
You do not need to name the skill. Describe your task in plain terms that match its scope. Requests that mention FASTQ files, counts matrices, STAR or Salmon quantification, nf-core/rnaseq configuration, or DESeq2 input are the ones it is built for.
If your data is single-cell or single-nucleus, say so. The skill's scope excludes those and points to the scanpy skill instead. If you only need the statistical fitting step, ask for pydeseq2 directly.
- 'Turn these FASTQ files and my Salmon output into a counts matrix for DESeq2.'
- 'Check my nf-core/rnaseq samplesheet for missing files and replicate problems.'
- 'Help me configure nf-core/rnaseq for GRCh38 with the STAR and Salmon aligner.'
- 'I have STAR output with featureCounts. Build a gene-level counts matrix and metadata template.'
Validate your sample sheet first
The skill recommends validating the sample sheet before any run, because this catches the most common failures early. The validator script checks the sheet for nf-core/rnaseq compatibility when given the nfcore option.
Ask the agent to run this check, or run it yourself from the skill folder. Fix every reported problem before starting the pipeline.
python scripts/validate_samplesheet.py --samplesheet samplesheet.csv --nfcoreChoose Path A or Path B for quantification
Path A uses nf-core/rnaseq, which runs FastQC, trimming, STAR or Salmon quantification, tximport and MultiQC with reviewed defaults. The skill recommends this path when you can use Nextflow and containers. Its gene counts come out already merged, so no bridge script is needed.
Path B runs each tool yourself. It suits a few samples, a constrained environment, or a step the pipeline does not expose. The skill warns that STAR/featureCounts and Salmon/tximport do not give interchangeable values, so choose one route for the whole comparison and quantify every sample the same way.
- Path A: field-standard, reproducible, scales to HPC or cloud
- Path B: transparent, good for learning and small projects
Build the counts matrix and hand off to DE
For Path B, the bridge script assembles a gene by sample counts matrix from Salmon, STAR or featureCounts output. It writes counts.csv, a metadata_template.csv to complete with your design (saved as metadata.csv), and counts_provenance.json. Keep your quantifier logs, tool versions and reference release alongside these files, because the provenance file cannot recover information missing from the quantification output.
Length-scaled Salmon counts are meant for full-length bulk RNA-seq. For 3′ counting assays, use original counts without transcript-length correction. Once the matrix and metadata are in place, pass them to the pydeseq2 skill for testing and to the pathway-enrichment skill for enrichment.
- counts.csv: integer counts, genes by samples
- metadata_template.csv: fill in the design, save as metadata.csv
- counts_provenance.json: input hashes, sample order, import settings
python scripts/build_counts_matrix.py --from salmon --quant-dir quant/ --tx2geneWhat people use it for
- Checking a samplesheet for missing FASTQ files and too few biological replicates before an nf-core/rnaseq run
- Assembling a gene-level counts matrix from Salmon quantification for a DESeq2 or PyDESeq2 comparison
- Configuring an nf-core/rnaseq run with a chosen genome and aligner and a pinned pipeline revision
- Teaching a student the FASTQ-to-counts workflow step by step, including where strandedness errors come from
Frequently asked questions
How do I install the bulk RNA-seq agent skill?
Download the ZIP from this page. Unzip it into ~/.claude/skills/ for personal use or into .claude/skills/ inside a project. In the Claude apps, upload the ZIP under Settings → Capabilities → Skills.
What does the bulk-rnaseq skill require to run?
It needs Python with pandas and numpy, and pytximport only for Salmon import. Path A needs Nextflow with a container engine, and Path B needs bioconda tools such as STAR, Salmon, fastp and subread. Network access is needed for installation and reference downloads.
Is the bulk-rnaseq skill free to use?
The skill is released under the MIT licence, as stated on this page. Check the licence file in the ZIP for the full terms. The tools it drives, such as nf-core/rnaseq and STAR, have their own licences.
Does the skill run differential expression?
No. It prepares the reads-to-counts stage and writes a counts matrix and metadata template. The statistical testing is done by the pydeseq2 skill or another DESeq2-based tool.
Can I use this skill for single-cell RNA-seq?
No. The skill covers bulk RNA-seq only. For single-cell or nuclei data, the skill points to the scanpy skill instead.
What is the difference between nf-core/rnaseq and the standalone route?
nf-core/rnaseq is a standard pipeline that runs QC, trimming, quantification and MultiQC with reviewed defaults. The standalone route runs each tool yourself, which suits small projects and step-by-step learning. The two can give different values, so pick one for the whole comparison.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you are preparing bulk RNA-seq data for differential expression and want a sample sheet and counts-matrix check before committing compute. Pick something else for single-cell or nuclei data, for clinical or diagnostic decisions, or if you expect a fully validated end-to-end run, since the full Nextflow, STAR and trimming recipes are illustrative and were not validated end to end.
Download the skill