AI-Designed Proteins: When Algorithms Learn to Fight Cancer
—
What if you could design a protein from scratch – one that doesn’t exist in nature – specifically engineered to hunt down cancer cells?
That’s not science fiction anymore. That’s what’s happening right now at the University of Washington.
—
The Baker Lab Revolution
Professor David Baker won the 2024 Nobel Prize in Chemistry for computational protein design. His team at UW’s Institute for Protein Design (IPD) has been using AI to do something remarkable: create entirely new proteins that nature never invented.
And in 2025, they’re putting those proteins to work against cancer and viruses.
—
Teaching Immune Cells to See Cancer
Here’s the problem with cancer: your immune system often can’t tell cancer cells from healthy cells. Tumors are sneaky – they’re made of your own cells gone wrong.
In July 2025, Baker’s team published research in Science showing AI-designed proteins that help immune cells recognize cancer markers. They created “binders” that lock onto specific proteins displayed on cancer cell surfaces – proteins called peptide-MHC complexes.
When they incorporated these designed proteins into CAR-T cell therapy (a treatment where a patient’s own immune cells are modified to attack cancer), the modified cells successfully identified and destroyed tumor cells in lab tests.
The target? A tumor antigen called PRAME, found in melanoma and other cancers.
The key insight: These proteins were designed entirely by AI. No natural template. No decades of trial and error. The algorithm figured out what shape would grab onto cancer markers, and the lab built it.
—
Cancer Drugs with an Off Switch
One of the scariest things about cancer immunotherapy is that it can work too well. Drugs like interleukin-2 (IL-2) supercharge your immune system to fight tumors, but they can also trigger dangerous inflammatory responses.
In September 2025, Baker’s team published research in Nature demonstrating proteins with built-in “on/off switches.” They engineered a controllable version of IL-2 that doctors could shut down rapidly if side effects became dangerous.
Think about that: a cancer drug you can turn off with a command. That’s the kind of precision control that could make aggressive treatments safer.
—
Antibodies Designed from Scratch
Antibodies are the workhorses of modern medicine – they’re behind treatments for everything from rheumatoid arthritis to COVID-19. The antibody drug market is worth over $200 billion.
Traditionally, developing a new antibody takes years. You inject an animal with a target, wait for its immune system to produce antibodies, then extract and modify the best ones.
In November 2025, Baker’s lab announced the first full-length antibodies designed entirely by AI. No animals. No trial and error. The computer designed them from scratch.
This could transform how we develop treatments. Diseases that were “undruggable” – because no natural antibody could target them – might suddenly become treatable.
—
Why This Matters Beyond Medicine
The techniques Baker’s team is developing aren’t just for cancer. AI protein design could create:
- Better enzymes for industrial processes (breaking down plastics, producing biofuels)
- New materials with properties we’ve never seen
- Targeted treatments for rare diseases that pharma companies ignore
- Sensors that detect specific molecules in the environment
We’re learning to program biology like we program computers.
—
The Makerspace Connection
You might wonder: what does this have to do with a makerspace in Anchorage?
A few things:
1. AI is becoming a maker tool
The same machine learning techniques that design proteins are becoming accessible. Tools like AlphaFold (protein structure prediction) are open source. The barrier between “cutting-edge research” and “stuff you can play with” is getting thinner.
2. Biotech is coming to makerspaces
Community bio labs already exist – places where people learn genetic engineering, grow bacteria, and experiment with biology. This is the next generation of making.
3. Understanding the science matters
When AI-designed cancer treatments hit the clinic, people will have questions. Is it safe? How does it work? Who controls it? The more people who understand the underlying technology, the better the public conversation.
4. Corny reads the right newsletters
Seriously. Staying curious about what’s happening in science – even fields far from your expertise – is a maker mindset. You never know when an idea from one domain will spark something in another.
—
Learn More
- Baker Lab: https://www.bakerlab.org/
- Institute for Protein Design: https://www.ipd.uw.edu/
- AlphaFold (open protein structure prediction): https://alphafold.ebi.ac.uk/
- Original GeekWire article: https://www.geekwire.com/2025/uw-researchers-create-ai-designed-proteins-to-hunt-down-cancer-and-viruses-plan-startup-spinoff/
—
This is the kind of future we’re building toward. Not just making things with our hands, but understanding how the world is being remade at every scale – from garage door openers to proteins that hunt cancer.
Thanks to Corny for sharing the article.
Recent Comments