Whether it's cancer, diabetes, or Alzheimer's disease, the proteins that acted badly are behind dozens of the most devastating illnesses.
If our body is an orchestra, the protein is our Maestros. As hormones, they cue processes such as growth, metabolism, and reproduction. As enzymes, they determine the tempo of chemical reactions required for digestion or DNA replication. As antibodies, they lead our immune system – and in yet other forms, they primarily side the life and death of cells.
However, if the protein is incorrect in shape or function, it causes disease.
Dr. Tanja Kortemme is Professor of Bioengineering and Associate Dean of Research at UC San Francisco, located in UCSF's pharmacy. Her lab recently created the world's first synthetic protein that changes shape. She tells us that UCSF scientists are using artificial intelligence (AI) to leverage decades of federal funding to create proteins they have never seen before. One day, these proteins may place their illicit brothers in their place.
How can protein be converted to a drug?
If a particular protein is not functioning properly in the human body, you want to use it as a drug. Take Insulin: Insulin is a hormone made from proteins that regulate metabolism, especially our body metabolizes sugar. However, in people with type 1 diabetes, their bodies do not produce enough insulin to metabolize sugar and control blood sugar levels, so insulin is prescribed.
Other protein-based drugs include GLP-1 weight loss drugs such as Ozempic and Wegovy, and antibody treatments such as Herceptin in cancer.
How can AI develop new drugs like this?
When you design a protein completely from scratch, you no longer restrict yourself to proteins that already exist. We can build proteins with completely new properties, which can be very powerful in solving the challenges we face in medicine.
The possibilities are almost limitless.
We are using AI to create new proteins from scratch. how?
There is an AI model that teaches you about protein structure. This means that all atoms within a protein are arranged and are three-dimensional. If your AI knows the protein structure, you can ask them to produce proteins that block different proteins that cause diseases such as cancer.
Where does the data that trains AI come from?
Many things can train AI models, but only succeed when training on a large database. For decades, a global community of researchers has been working on determining protein structures. To cooperate with this work, scientists deposited their findings into the Open Access Protein Data Bank. This database was made possible with federal research funding from the National Institutes of Health (NIH) and the National Science Foundation.
My lab uses large-scale data from banks and other sources to generate novel proteins with new features. We do this by using existing generated AI models and creating our own new AI models.
How do proteins move from AI models to real life?
This is possible through DNA synthesis and recombinant DNA techniques that allow scientists to construct DNA segments to produce proteins.
Recombinant DNA technology is one of the major innovations in UCSF.
Are there any AI-designed protein drugs already on the market?
Many companies use AI methods to help them discover and optimize potential drugs. But are purely AI-designed proteins from scratch? I don't have it yet.
Still, there is an explosion of efforts in the biotech industry to develop AI-producing proteins with therapeutic implications. We would expect many of these designed proteins to enter preclinical development over the next five years and hopefully enter the clinic to actually help people.
Is UCSF at the forefront of AI-designed proteins?
Yes, for over 20 years, UCSF has had a lot of strength in pioneering protein engineering techniques to construct proteins in older ways to transform existing proteins into better medicines in more classical ways. Today, the UCSF group – especially our incredibly innovative graduate students and postdoctoral scholars, have developed advanced computational methods, including AI, to design proteins from scratch.
What is the secret to this kind of innovation?
Much of UCSF's advancements are through interdisciplinary science and collaboration, creating opportunities for innovation when different disciplines come together. And our postdocs and graduate students are a big part of that.
Many students applying to UCSF are interested in artificial intelligence. Often they have a background in computer science, engineering, or mathematics. This is all areas that provide the right background to advance AI. However, these students are also fascinated by the biological and biomedical issues that drive research at UCSF.
UCSF provides an incredible environment for integrating basic biology, discovery and biomedical sciences, and pharmaceutical development, as well as computer science and engineering and other disciplines.
