Hello Bioinformatics lovers,

Tommy here. It is the Chinese Mid-autum festival today during which the families get together and eat the moon cakes.

Travel safely if you are in the areas that the huge Nor'easter storm hits. I am staying home because of the heavy rain and strong wind.

Now, let's get into today's newsletter.

We over-expressed a wild-type protein that belongs to a multi-subunit complex, expecting it to restore the complex's function. As a control, we also over-expressed a non-functional version of the same protein. I assumed the broken protein would do nothing.

It did something. The complex lost function.

The result made no sense to me. The wet lab scientists who ran the experiment explained it instantly: a dominant-negative effect.

Why a dead protein can break a working complex

The mutant protein still binds its partners. It just can't do its job once it's inside the complex.

When you over-express it, the mutant floods the cell and takes the slots that the wild-type protein should fill.

Most of the complexes that assemble now carry a defective subunit, and overall function drops.

Once I had the name for it, the results made sense.

What a year of daily Claude Code taught me

I have used Claude Code almost every day for more than a year of bioinformatics work. This story sums up three lessons.

1. Writing code stopped being the hard part. Framing the right question and giving Claude Code clear instructions took its place.
   ​
2. Judging the output is the real bottleneck. I still spend a lot of time reading the logic, reworking my question, and asking Claude Code to rewrite the code. Claude Code could write every line of the analysis for this experiment. It took a biologist to explain what the numbers meant.
   ​
3. Your collaboration with the wet lab is your edge. The scientists who generate the data carry biology you may never find in a code review. Clear, frequent conversations with them turned a confusing plot into a textbook mechanism.

Try this

The next time a result surprises you, take it to the scientist who generated the data before you ask AI to explain it. Bring the plot and the experimental design. Ask what they expected to see.

AI is changing how we do bioinformatics. It is still a tool, and you need to be the pilot.

Hit reply and tell me: when did a piece of biology rescue one of your analyses? I read every response

Happy Learning!
Tommy aka crazyhottommy

PS:

if you find the newsletter helpful, please forward the email to your friends :)

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!