Scientists at the joint School of Engineering and School of Life Sciences Laboratory of Protein Design and Immunoengineering (LPDI) have developed MaSIF, a machine learning-powered method for analyzing and mapping protein surface structures and determining their functional properties faster than ever before. After years of research and development, this team led by Bruno Correia has been able to engineer novel proteins that can interact with a range of therapeutically-relevant targets such as the SARS-CoV-2 spike protein.
Molecular interactions between proteins govern a huge range of biological activities, but these connections can be hard to predict due to the vast diversity of protein surfaces. MaSIF was developed to bridge this gap. The machine learning system processes protein surface fingerprints and identifies complementary surfaces for specific protein target sites. Fragments of these surfaces are then digitally grafted onto larger protein scaffolds, with the strongest-binding binders then isolated and synthesized in the lab.
When the team started applying their process to therapeutically-relevant protein targets, they found they could generate accurate high-affinity binders in just a couple of months. This breakneck speed opens the door to rapidly programmed epidemiological responses and opens new possibilities for CAR-T therapies and the development of novel protein-based drugs.
By leveraging the power of machine learning, the LPDI’s process has potentially revolutionized protein engineering. Their work provides an empowering example of how computational models and modern data science principles can be used to devise creative solutions to complex scientific challenges.
The LPDI is a joint collaboration between the School of Engineering and School of Life Sciences located at the University of California, Santa Cruz. It is dedicated to advancing experimental platforms, software and machine learning approaches for protein design, engineering and immunoengineering. The head of the lab, Bruno Correia, is a leader in computational biology, with extensive experience in applying computer modeling to the study of protein structure and function. He is also well-published in the area of synthetic biology and protein engineering.
This success of this research team from the LPDI demonstrates the immense potential of combining machine learning and data science principles for purposes such as discovering novel proteins, engineering drug delivery systems and advancing evidence-based therapies. It also illustrates how digital modelling can be successfully applied to harness the powerful and dynamic capabilities of protein-based systems.
