When to Use ChatGPT for Bioinformatics (And When NOT To)

Hi everyone! This video is about possible use cases for ChatGPT in bioinformatics, as well as warnings for when to avoid it, mostly based on my experience at my regular day-job and when I've most found it to be helpful. Thanks for watching and let me know if you have any questions!

[@CatlinZenith]

Mike, I am trying to use gpt 4.0 to learn the workflow of ATAC-seq and ChIP-seq. Have you tried using things in the gpt store to resolve the issue instead of using gpt3.5? im wondering its much convincing when it comes to 4.0

[@MikeSaintAntoine]

Hi Catlin! That's a good question, but no I haven't tried any of the specialized models from the GPT store, just GPT 4. In my experience so far, GPT4 has been pretty good for getting a broad overview of different types of sequencing experiments, but I still wouldn't 100% trust it yet on technical questions. Basically my rule of thumb for using ChatGPT in research is to only use it when I can verify myself that it's definitely getting it right (so for things like plotting, string formatting, regular expressions, etc) or to use it as a starting point to get an overview of a topic, before going on to eventually read actual research papers about the topic.

[@CatlinZenith]

@@MikeSaintAntoine You're absolutely right, I've started using it more as a traditional language model should be used. For instance, I've been working with the nf-core cutandtag pipeline recently, and I've tasked it with processing some documentation for discussion. It's made my interactions with the developer team on Slack much smoother, and debugging is incredibly quick.

[@shayanakhavan6002]

Great video, Mike! I like how you call GPT providing misinformation to you as a hallucination 😂

[@MikeSaintAntoine]

Ha! Yeah that's what I've been hearing people call it. I guess it's better than just calling it BS lol