Hello everyone, Last year I wrote several posts on the large language models (LLM) of DNA sequences (evo/evo2 papers), and some readers emailed me asking about the plans to continue the series. Here is an exciting update for you. Guhan Thiagarajan, a high-school student, worked with me and built a “toy model” to help you explore these concepts.
In earlier blog posts, we explored how AI models are being used for protein folding and DNA language modeling. Machine learning is driving rapid progress in another important area of biology: antibiotic discovery. Rushil Mohan, a smart high-school student, worked with me and wrote a fantastic review post to introduce you to this fascinating area of research.
On Saturday, I ran Alphafold for the first time. Google built a website (Alphafold Server) through which you can run the program directly. The results are really positive, or cross that, extremely positive (wink). Here I provide a brief tutorial and a description of what I found.
Note: Please check our full plan for the series here
Alright class ! If you are still here, you already completed part (i) covering Scikit-learn for machine learning and became familiar with the Pytorch library for AI in part (ii). As part of the first lesson, you also watched the StatQuest videos explaining four machine learning concepts.
Note: Please check our full plan for the series here
In part (i) of this guide, you learned to use Python library Scikit-learn for machine learning. Today you will get familiar with Pytorch for AI.
Note: Please check our full plan for the series here
These days, many biologists are trying to learn machine learning and artificial intelligence. That is not surprising. Recent research papers in genomics, molecular biology, drug discovery, ecology, or bioinformatics are increasingly using terms and concepts from those fields. Here is a short roadmap on how to start, and moreover, how to proceed rapidly through the maze.
In a classic scene from Idiocracy, we see the citizens of a fictional United States from year 2500 repeatedly insisting that Brawndo is the “thirst mutilator” because it contains electrolytes, while water belongs only in the toilet. The slogan is repeated so many times by the Brawndo company that it has become deeply embedded in their thinking.
Right now everyone is either enamored of or is completely turned off by AI. Speaking of being fed up, please check the students at University of Florida booing at this commencement event. Much of their reactions comes from intense hype generated by the tech-bros of Silly Con valley to increase stock prices of their companies.
In my earlier post, I mentioned about three distinct activities described under the broad term “AI”. They are -
(i) using web-based text engines like Chatgpt, Claude or Gemini and their extensions as coding tools,
(ii) downloading numerical models directly from Huggingface and building applications on top of them, and
(iii) developing and training mathematical models for new applications.
In 2011, I wrote two articles (here and here ) providing beginners’ guides to bioinformatics. Eight years later (2019), I posted an updated guide here. Now that AI has become a powerful tool, it is time to discuss how the work of bioinformatics and computational biology is changing.
I have been coding with AI assistance for about six months now and more actively since late December 2025. If I had to compress the entire experience into a single sentence, it would be this: AI made me faster first and then it made me slower. Let me explain.
In the earlier posts of this series (here, here, here and here), we covered the mathematical and biological aspects of evo and evo2. One important topic that we have not covered yet is how the models were trained.
In this article, I will argue that Multi Parameter Statistics, or even better, Massively Parameterized Statistics (MPS) better describes the application of AI models in biology and medicine. Also, I will introduce you to a new preprint on DNA sequence modeling that claims to match evo.
In the last three posts of this series (here, here and here), we covered the mathematical aspect of evo and evo2. Let us now discuss the biological findings from these models. It will take multiple posts to go over these topics.
In the first two posts of this series (here and here), we covered the AI-related mathematical concepts applied to evo and evo2. Before moving on to the biological side, here is one last post on the model.
In the first post of this series, we covered the basic technical terms of the evo and evo2 papers. We also mentioned the key technological innovation that made their work possible. That led to the question - if they were using fast fourier transform (FFT), were they using convolutional neural network (CNN)? The answer is no. The computer science work done by the Stanford group is quite groundbreaking. Let me go over that in detail.
