Where do I think the next amazing revolution is gonna come? And this is gonna be flat out one of the biggest ones ever.
There’s no question that digital biology is going to be it.
For the very first time in our history, in human history, biology has the opportunity to be engineering, not science. When something becomes engineering, not science, it becomes less sporadic and exponentially improving. It can compound on the benefits of the previous years. And every researcher’s contributions compound on each other. For the very first time, we know how to represent biology, understand the language of biology, we can represent the language of chemistry. This never happened before. I’m very proud to say that Nvidia is the center of all of that, and we’ve made it possible for some of those breakthroughs to happen. And now we’re gonna have incredible tools that bring the world of biology which is very chaotic and random, and constantly changing, diverse and complex, and bring it into the world of computer science. And that is going to be profound. And if you happen to love this intersection, I think it’s going to be rich with opportunities. It’s going to be a giant industry.
-Jensen Huang, Founder, CEO, Nvidia Corp.
Subscribe to get regular posts discussing the projects I’m working on.
Background
At the beginning of my senior year of undergrad, I signed up to take a computational biology course. This course was my first real exposure to writing software. Before then, the code I wrote was either basic Python scripts that were purely an exercise in getting familiar with programming or MATLAB code for modeling a biological system. In the biomedical engineering program (it’s been 5 years since I graduated, so curricula might have changed), you don’t learn how to build production-grade software.
But the comp bio course was the best way to learn software development because it combined my emerging interest in computing with my major, biomedical engineering. I learned a lot of useful skills that I still use today, notably writing shell scripts and SSH’ing into a computing cluster to run molecular dynamics simulations.
After 3 years of writing software professionally in a field not related to biology at all, I’m finally circling back to my original field of study. But as I become reacquainted with the intersection of biology and engineering, I find that now there is a much greater emphasis on software and, of course, artificial intelligence. The field is now increasingly being named as “TechBio.”
What exactly do we mean by “TechBio”?
First, here’s a quick primer on the relevant biology, but you may need to draw on high school biology you haven’t thought about in years.
Central dogma of Biology
Protein engineering1
To summarize, we can change a protein’s function by changing the sequence of amino acids that comprise the protein. By harnessing our relatively recently gained ability to edit DNA in organisms, we can compose new proteins. But this is easier said than done, because we still don’t know how a protein sequence will “fold”, which is the process of stabilizing into a three-dimensional structure.
Defining TechBio
I found a great definition for TechBio in a Bloomberg News article:
Kimberly Powell, head of Nvidia’s health-care segment, says the pharma business is quickly changing in response to AI, with many of the startups Nvidia works with today considering themselves “techbio” companies rather than biotechs—using data to “drive what biology they’re going after, instead of the biology influencing what technology they need to use.”2
Additionally, Elliot Hershberg writes in his newsletter, The Century of Bio:
An important transition is also taking place in the world of biotech. From the headline news of Google DeepMind solving protein structure prediction, to the recent completion of the human genome, the frontier of biology has become deeply integrated with AI, software, and hardware. The new generation of hybrid companies on the cutting edge of this trend are often referred to as techbio companies.1
This new intersection between bits and atoms is a really big deal. In a world where the cost of developing new drugs is increasing exponentially and healthcare takes up an unsustainable portion of our GDP, something needs to change. Beyond developing more medicines cheaper and faster, techbio could help us live longer and healthier lives in material abundance made possible through the biologization of our industrial processes.3
I would summarize it as combining biology and engineering by utilizing our ability to read/write DNA and layering on software to create useful biological systems.
Applications
I categorize the applications of TechBio into two categories: industrial and clinical.
Industrial
Computational tools for designing biological systems for:
Biofuel production
Agriculture
Materials
Chemicals
Robotics + Lab automation for accelerating the design/build/test/learn cycle
Scalable bioreactor platforms
Clinical
AI for drug discovery
Precision medicine
Digital twins for clinical trials
Digital health
Clinical applications are typically characterized by long development times and FDA regulatory pathways.
My undergrad coursework and lab research were focused only on clinical, so I want to try something new and explore the industrial applications, but I’m sure I’ll circle back to interesting clinical opportunities someday.
Upcoming Projects
I plan to write a series of articles titled “Coding Cells” that will cover my projects and learnings on synthetic biology, computational biology, bioinformatics, and bioengineering.
Here are the topics I’m interested in:
AlphaFold
Machine learning model developed by DeepMind, a Google subsidiary, that predicts the three-dimensional protein structure from an amino acid sequence.
The model execution timed out for me when running the Jupyter notebook on Google Colab, so I’m working on building an app to run ColabFold on my own infrastructure on Google Cloud.
I’m planning on adding visualization and docking features to the web application. I plan on discussing how I built this in future articles.
ColabFold: open-source model
Bioreactors
I bought a Pioreactor, which is a modular open-source bioreactor that I came across on Hackernews. As of Feb 2024, a Pioreactor costs $249.00, which is not cheap, but I don’t think this can be helped since the company is likely not manufacturing at a big enough scale to offer a low unit price to customers.
I’m interested in building a software layer on top of modular bioreactors and exploring the integration of machine learning, primarily for directed evolution. I’ll discuss my bioreactor experiments in future articles.
Lab Automation and Services
Lab robots for automation
For example, automating a lab protocol by scripting a pipetting processing using Python.
Sequencing services
Helpful for people like me who don’t have in-house sequencing equipment
Service providers: Primorodium, Plasmidsaurus
Cloud Labs
Access lab equipment remotely to conduct your experiments.
Interesting trend that could have similar implications as cloud computing did for computing and information technology.
Resources
To conclude, here’s a collection of resources to learn more about the opportunities and state-of-the-art in TechBio:
The Century of Biology - Eliot Hershberg
Asimov Press - Niko Mccarty
a16z Raising Health podcast
How to Grow Almost Anything - MIT
I’m currently taking this course. It’s a free online course focused on synthetic biology - both theory and lab skills.
Thanks for reading, and leave a comment if you have any feedback. Feel free to share this article with whomever you think would be interested.
Special thanks to Sravya Varanasi for proofreading drafts of this article.
Interested in receiving this newsletter directly to your inbox?
Think you know someone who might be interested in reading this article?
Prof. Jeffrey Gray, Computational Protein Structure Prediction and Design, Johns Hopkins University
AI Can Speed Drug Discovery. But Is It Really Better Than a Human?, Bloomberg News, 2024
Elliot Hershberg, TechBio Taxonomy, 2022