Researchers develop AI model to predict protein-DNA binding accuracy

Researchers at the University of Southern California have unveiled a groundbreaking AI model designed to predict the accuracy of protein-DNA binding. Named Deep Predictor of Binding Specificity (DeepPBS), this innovative tool aims to expedite the development of new drugs and medical treatments by providing more efficient predictions of how proteins interact with DNA.

DeepPBS is built on a geometric deep learning framework that analyzes the structures of protein-DNA complexes to forecast binding specificity. This approach allows scientists to input data on these complexes into an online computational tool, thereby streamlining the process of understanding protein-DNA interactions. According to Remo Rohs, a Professor at USC’s Dornsife College of Letters, Arts, and Sciences, DeepPBS eliminates the need for time-consuming high-throughput sequencing or extensive structural biology experiments, making it a faster and more efficient method for determining binding specificity.

What sets DeepPBS apart from existing methods is its ability to predict interactions across various protein families, rather than being limited to a specific protein family. This universal approach is crucial for researchers who need a versatile tool for studying diverse proteins and designing new ones. DeepPBS captures both the chemical properties and geometric contexts of protein-DNA interactions, providing a comprehensive analysis that enhances the accuracy of predictions.

Building upon advances in protein-structure prediction, such as DeepMind’s AlphaFold, DeepPBS complements these technologies by predicting binding specificity even for proteins that lack available experimental structures. This capability could significantly accelerate the design of new drugs, improve treatments for genetic mutations and cancers, and drive innovations in synthetic biology and RNA research.

This development marks a significant advancement from the seminal discovery of DNA by Francis Crick and James Watson in 1953, illustrating how far our understanding and technological capabilities have come in the field of molecular biology.