How Artificial Intelligence Is Advancing Structural Proteomics

This article is based on research findings that have not yet been reviewed. Therefore, the results are considered preliminary and should be interpreted as such. Learn about the role of the peer review process in research over here. For more information, please contact the mentioned source.

Understanding protein complex composition is critical in drug design and development of therapeutic proteins such as antibodies. However, proteins can bind to each other in millions of different combinations, and the current docking solutions used to predict these interactions can be very slow. Faster and more accurate solutions are needed to simplify the process.

in Prepress Published earlier this year, a new machine-learning model – EquiDock – can quickly predict how two proteins will interact. Unlike other approaches, the model is not based on dense filter samples and has been shown to reach predictions up to 80 to 500 times faster than common docking programs.

To learn more about EquiDock and how artificial intelligence (AI) methods are advancing the field of structural proteins, technology networks Talk to the paper’s co-lead author, Octavian Eugene Janiais a postdoctoral researcher in the Computer Science and Artificial Intelligence Laboratory at the Massachusetts Institute of Technology.

Molly Campbell (MC): For our readers who may be unfamiliar, please describe your current research focus in proteins?

Octavian Jania (OG): My research uses artificial intelligence (specifically, deep learning) to model aspects of molecules that are important in various applications such as drug discovery.

Proteins are involved in most biological processes in our bodies. Two or more types of proteins with different functions interact and form larger machines, such as complexes. They also bind to smaller molecules such as those found in medicines. These processes alter the biological functions of individual proteins, for example, an ideal drug would inhibit a cancer-causing protein by binding to certain parts of its surface. I am interested in using deep learning to model these reactions and to help and accelerate the research of chemists and biologists by providing better and faster computational tools.

MC: How are AI-based approaches advancing the field of structural proteomics and proteomics specifically?

OG: Biological processes are very complex in nature and have their own secrets, even for experts in the field. For example, to understand how interacting proteins relate to each other, humans or computers have to try all possible attachment combinations in order to find the most plausible one. Intuitively, having two 3D objects with very irregular surfaces, one must rotate them and try to fit them in every possible way until one can find two complementary regions on both surfaces that match well in terms of geometric and chemical patterns. . This is a time consuming process for both manual and computational methods. Moreover, biologists are interested in discovering new interactions across a very wide range of proteins such as the human protein which is about 20 thousand in size. This is important, for example, to automatically detect unexpected side effects of new treatments. Such a problem is now similar to a very large 3D puzzle where one has to scan identical pieces at the same time and also understand how each binary attachment occurs by trying all possible combinations and courses.

MC: Can you explain how the EquiDock was created?

OG: EquiDock takes the 3D structures of two proteins and directly identifies the regions likely to interact which would be a complex problem even for a biologist. Discovering this information is then sufficient to understand how the two proteins are rotated and directed at their related positions. EquiDock learns to capture complex docking patterns from a large set of approximately 41,000 protein structures using a geometrically constrained model with thousands of parameters that are dynamically and automatically tuned until the task is very well solved.

MC: What are the potential applications for EquiDock?

OG: As mentioned earlier, EquiDock can enable fast computational scanning of drug side effects. This is in line with the hypothetical large-scale screening of drugs and other types of molecules (for example, antibodies, nanobodies, and peptides). This is necessary in order to drastically reduce the astronomical research space which would be useless for all of our current experimental capabilities (even aggregated worldwide). A rapid protein docking method such as EquiDock with a rapid protein structure prediction model (such as AlphaFold2 developed by DeepMind) will help drug design, protein engineering, antibody generation, or understanding drug action, among many other exciting applications much needed in our search for therapeutics. Better for disease.

Octavian Janea was talking to Molly Campbell, the first science writer for Technology Networks.

Leave a Comment