AI Framework 'MULTI-evolve' Accelerates Protein Engineering
A new machine learning framework called MULTI-evolve enables researchers to explore protein "sequence space" orders of magnitude faster than traditional methods. The tool allows for the design and prediction of properties for millions of protein variants, aiming to shorten discovery timelines in biotech product development and drug design.
- The MULTI-evolve framework was developed by a team at the Arc Institute, led by bioengineer Patrick Hsu and researcher Vincent Tran. Their method compresses the traditionally lengthy, iterative process of protein engineering into a single experimental round that can be completed in weeks. - A key innovation of this framework is its focus on "epistasis," or the interaction effects between different mutations. By measuring how pairs of mutations work together, the model can more accurately predict the function of proteins with multiple simultaneous mutations. - In one test case, MULTI-evolve was used to engineer an APEX peroxidase enzyme, resulting in a variant with a 256-fold increase in activity. It also improved the binding affinity of an anti-CD122 antibody and enhanced the function of a dCasRx protein used for RNA trans-splicing by nearly tenfold. - This type of work is central to the field of bioinformatics, which combines biology, computer science, and statistics. Professionals in this area, often called bioinformaticians or computational biologists, write code (using languages like Python and R) to analyze large biological datasets and develop algorithms to interpret them. - A career in bioinformatics or computational biology typically requires at least a bachelor's degree with foundational courses in biology, computer science, math, and chemistry. Many roles, especially in research and academia, require a master's or even a Ph.D. - The day-to-day work in computational roles is primarily computer-based and involves analyzing data, writing and debugging code, running analyses on high-performance computing clusters, and collaborating with lab-based scientists to interpret results. This contrasts with patient-facing roles, which are centered on direct interaction with patients, or lab research roles that involve hands-on experiments. - The tools and approaches developed in bioinformatics, like MULTI-evolve, directly impact patient care by accelerating drug discovery and the development of new therapeutics, such as monoclonal antibodies for treating cancer or autoimmune diseases. This creates a bridge between tech-focused and patient-facing careers in life sciences.
