Hello Bioinformatics lovers,

Tommy here.

Hadley Wickham's ((btw, you should know tidyverse by him) colleague built a benchmark that flips the relationship hidden in the data behind a fuel-economy plot.

On screen, engine size climbs and miles per gallon climbs with it, a pattern no one who has driven a car would believe.

About 80% of the time, large language models describe the plot as showing the opposite. They lean on the axis labels and the relationship they expected, and never look hard at the trend in front of them..

Wickham just published a piece called "y code when ai?" He built the tidyverse and ggplot2. He now leans on Claude for roughly 90% of the R code he writes (me too!). He still argues that code, and the judgment behind it, matters more than it ever has.

His framing is that AI works as an amplifier.

Feed it a shallow understanding of your problem and it produces mediocre work faster than you could alone.

Feed it real expertise and you can do the best work of your career.

The model grinds through its own feedback loop, writing code, running it, reading the error, trying again. It cannot tell you whether a plot is informative or a model fits the biology. You own that call.

The lazy plot reader is our problem too. A bioinformatics pipeline runs clean and hands back a coordinate off by one, a swapped cluster label, a gene ID that never existed in the reference. The tools stay silent. You catch it because you understand the assay. The code never raises a flag.

So the expertise has to come from somewhere. Here are ten courses I keep pointing people to. Several come from the people who built the tools you run every day:

1. ​Data Science / HarvardX PH525x by Rafael Irizarry
2. ​Applied Computational Genomics by Aaron Quinlan, author of bedtools
3. ​Bioinformatics Algorithms by Phillip Compeau and Pavel Pevzner, with the video course on Coursera
4. ​The Biostar Handbook by Istvan Albert, founder of biostars.org, taught at Penn State
5. ​Introduction to Bioinformatics and Computational Biology by X. Shirley Liu
6. ​Data Carpentry Genomics Workshop​
7. ​Applied Comparative Genomics by Michael Schatz
8. ​Introduction to Computational Biology by Mike Love, creator of DESeq2
9. ​MIT Computational Biology 6.047 by Manolis Kellis
10. ​An Introduction to Applied Bioinformatics by Greg Caporaso, with Python code

I wrote the ChIP-seq chapter in the Biostar Handbook and had a hand in a couple of the others, so I have watched people learn from them for years. The people who understand their data get the real leverage. AI amplifies what is already there.

Which course did you start with, or which one has you stuck right now? Hit reply and tell me. Your answer shapes what I write next.

Happy Learning!

Tommy aka crazyhottommy

PS:

If you want to learn Bioinformatics, there are four 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. Lastly, I post daily on Linkedin https://www.linkedin.com/in/%F0%9F%8E%AF-ming-tommy-tang-40650014/recent-activity/all/​

Stay awesome!