Hello Bioinformatics lovers,

Tommy here.

You know that moment when Claude suggests using FPKM for your RNA-seq analysis?

Or writes a Snakemake rule with the wrong parameter syntax and you spend 20 minutes debugging it?

I kept hitting this. Claude would generate code that looked right but failed on basic bioinformatics conventions.

Wrong tool versions. Outdated best practices. Parameters that hadn’t been recommended since 2015.

Then I discovered Claude Skills, and everything changed.

What’s a Skill?

It’s just a markdown file. That’s it.

You write down your instructions—”when analyzing RNA-seq data, always check for strandedness”—and Claude remembers.

Like handing a new lab member your protocol notebook instead of hoping they guess the right method.

No coding required. No complicated setup. Just create a .md file with your guidelines.

The difference in practice:

Before: “Write a Snakemake rule for RNA-seq quantification”
→ Wrong parameter syntax
→ 20 minutes of debugging
→ Frustration

After (with Skills loaded): Same prompt
→ Correct syntax first try
→ Proper output format
→ Actually saves time

Two repos that make this dead simple:

1. ​bioSkills - Computational biology workflows
2. ​claude-scientific-skills - Full research lifecycle coverage

Both are collections of markdown files teaching Claude our field’s best practices.

What’s actually in these files?

Here’s a snippet from the bioSkills RNA quantification skill:

“Quantify transcript abundance directly from FASTQ reads using pseudo-alignment (kallisto) or selective alignment (Salmon).”

Then it provides example commands with flag explanations.

Simple instructions. Massive impact.

How to use them:

In Claude Code: /plugin install scientific-skills@claude-scientific-skills

Or download the markdown files and drop them in your .claude/skills folder.

That’s it. The files teach Claude what good bioinformatics looks like. Pick what you need. Load multiple skills simultaneously.

Write your own:

Got a custom lab workflow? Document it:

​
When analyzing our ChIP-seq data:
- Use hg38 reference (we switched in 2023)
- Filter blacklist regions from ENCODE
- Call peaks with q-value < 0.05
​

Save as SKILL.md. Done.

The bottom line:

Your prompts stay the same. The output gets dramatically better.

If Claude’s bioinformatics suggestions have been frustrating you, try loading these community skills.

It’s like the difference between giving someone vague directions versus handing them a GPS with your exact route programmed in.

The tools finally work the way we hoped they would.

What’s your biggest frustration with AI coding tools? Hit reply and let me know—I read every response.

Happy Learning!

Tommy aka crazyhottommy

PS:

If you want to learn Bioinformatics, there are other ways that I can help:

1. My free YouTube Chatomics channel, make sure you subscribe to it.
2. I have many resources collected on my github here.
3. I have been writing blog posts for over 10 years https://divingintogeneticsandgenomics.com/​
4. Read all my last week's Linkedin posts https://www.linkedin.com/in/%F0%9F%8E%AF-ming-tommy-tang-40650014/recent-activity/all/​

​

Stay awesome!