Getting into BioML from an ML background can be pretty intimidating. For me, taking some time to learn the fundamentals of molecular biology first helped make the problems in the field feel more approachable. It also turned out to be incredibly beautiful and interesting, reinforcing my motivation to dig deeper.
"The Machinery of Life" [1] was fantastic for understanding the molecular and cellular world. The author is a scientist and an artist, and his oil paintings do an incredible job of illustrating that world. I kept wishing I'd seen such images in my high school textbooks. If I had, I might’ve ended up in this field much earlier.
After that, "Quickstart Molecular Biology" [2] reiterated some of the core concepts, but more importantly, it helped connect those basics to the kinds of data and techniques commonly used in BioML. Note: some people recommended starting with this one, but I found it to be "too much too soon" without reading "The Machinery of Life" first.
I’m still early in the journey, but I’m really grateful to the folks on Reddit who recommended these books, so I wanted to pass them along in case they help someone else starting down this path.
The emphasis on engineering discipline is spot-on. I didn't see it in the post so I'm adding here: As a whole, BioML is trying to systematize research in a field that has been mostly un-engineerable since its conception, and as a result has very few paradigms beyond evolution and the central dogma. Domain knowledge is siloed and very stratified, so rapid and messy early prototyping is important to break down the barriers to entry. But if you want people to pay attention to your work, you need to make tools. Methods are the currency of biology (the most cited bio papers of all time are mostly methods [1]), and computational work is only reinforcing this. Building tools is the ultimate test of your effficacy in BioML because (1) you have to think deeply about how others are going to use your work (2) good tools should be easy to use and hide complexity; people shouldn't have to understand how you built a tool to use it, and (3) you have to recognize where your research is going in order to make investing in tools worthwhile for yourself long-term. 3 is the hardest, 1 and 2 usually follow in order if you get it right. The best researchers and engineers I know aren't just clever, but they build their own virtual toolboxes to carry with them everywhere they go.
Given your interest in Biotech/AI, you might enjoy my recent piece on eXoZymes Inc. They’ve just commercially launched a cell-free enzyme biocatalysis platform that converts biofeedstocks into targeted chemical products and the platform heavily integrates AI/Computational models to design and synthesise its compounds in mere weeks.
Plus they just announced their first subsidary which synthesises N-trans-caffeoyltyramine (NCT) to treat MASLD/MASH. Very very interesting compound that has immense potential and went from idea to initial syntheise in just 6 weeks with the integration of AI
Getting into BioML from an ML background can be pretty intimidating. For me, taking some time to learn the fundamentals of molecular biology first helped make the problems in the field feel more approachable. It also turned out to be incredibly beautiful and interesting, reinforcing my motivation to dig deeper.
"The Machinery of Life" [1] was fantastic for understanding the molecular and cellular world. The author is a scientist and an artist, and his oil paintings do an incredible job of illustrating that world. I kept wishing I'd seen such images in my high school textbooks. If I had, I might’ve ended up in this field much earlier.
After that, "Quickstart Molecular Biology" [2] reiterated some of the core concepts, but more importantly, it helped connect those basics to the kinds of data and techniques commonly used in BioML. Note: some people recommended starting with this one, but I found it to be "too much too soon" without reading "The Machinery of Life" first.
I’m still early in the journey, but I’m really grateful to the folks on Reddit who recommended these books, so I wanted to pass them along in case they help someone else starting down this path.
[1] https://www.amazon.com/Machinery-Life-David-S-Goodsell/dp/0387849246
[2] https://www.amazon.com/Quickstart-Molecular-Biology-Introductory-Mathematicians/dp/1621820343
The emphasis on engineering discipline is spot-on. I didn't see it in the post so I'm adding here: As a whole, BioML is trying to systematize research in a field that has been mostly un-engineerable since its conception, and as a result has very few paradigms beyond evolution and the central dogma. Domain knowledge is siloed and very stratified, so rapid and messy early prototyping is important to break down the barriers to entry. But if you want people to pay attention to your work, you need to make tools. Methods are the currency of biology (the most cited bio papers of all time are mostly methods [1]), and computational work is only reinforcing this. Building tools is the ultimate test of your effficacy in BioML because (1) you have to think deeply about how others are going to use your work (2) good tools should be easy to use and hide complexity; people shouldn't have to understand how you built a tool to use it, and (3) you have to recognize where your research is going in order to make investing in tools worthwhile for yourself long-term. 3 is the hardest, 1 and 2 usually follow in order if you get it right. The best researchers and engineers I know aren't just clever, but they build their own virtual toolboxes to carry with them everywhere they go.
[1] https://www.nature.com/articles/d41586-025-01124-w
Hey Nathan
Given your interest in Biotech/AI, you might enjoy my recent piece on eXoZymes Inc. They’ve just commercially launched a cell-free enzyme biocatalysis platform that converts biofeedstocks into targeted chemical products and the platform heavily integrates AI/Computational models to design and synthesise its compounds in mere weeks.
Plus they just announced their first subsidary which synthesises N-trans-caffeoyltyramine (NCT) to treat MASLD/MASH. Very very interesting compound that has immense potential and went from idea to initial syntheise in just 6 weeks with the integration of AI
https://www.slack-capital.com/p/exozymes-research-report