Researchers at The University of Texas at Dallas have developed a novel machine-learning model named ProSSpeC, designed to predict how protease enzymes function. This breakthrough addresses a significant bottleneck in drug development, where the inability to accurately forecast protease behavior has slowed the creation of treatments for viruses and cancer. Published in Nature Communications, the study details an interdisciplinary approach that combines bioengineering, evolutionary biology, and computational analysis to design more precise molecular tools. Proteases act as molecular scissors within the body, cleaving proteins to stop viral replication or trigger cell death in cancer. However, creating medicines that target these enzymes has traditionally relied on slow, labor-intensive trial-and-error processes in the laboratory. ProSSpeC changes this paradigm by analyzing evolutionary data spanning millions of years. The model examines the phylogeny of the Potyviridae family of plant viruses, comparing thousands of related enzymes with different amino-acid sequences. By identifying which parts of a protease are essential for its function and which can be altered, the system predicts whether a specific engineered variant will successfully cleave a target protein substrate. Dr. P.C. Dave Dingal, an assistant professor of bioengineering, and Dr. Faruck Morcos, an associate professor of biological sciences, led the team. They stated that the model essentially learns from evolutionary history, allowing researchers to determine which engineered variants will function without testing thousands of random possibilities in the lab. This method mimics nature's guidance in building superior molecular tools, effectively bridging the gap between computational prediction and experimental validation. In their experimental validation, the team produced new proteases based on ProSSpeC's suggestions and tested their efficacy. The results showed that the model successfully identified synthetic proteases that outperformed the tobacco etch virus protease, a standard enzyme widely used for protein purification in both research labs and pharmaceutical production. The researchers have filed a provisional patent for these highly effective enzymes. Medel B. Lim Suan Jr., a biomedical engineering doctoral student and co-first author of the study, highlighted the value of this interdisciplinary work. He noted the unique opportunity to gain experience by integrating computational modeling with hands-on experimental biology, observing how the two fields translate into one another. The team believes this approach opens new avenues for designing enzymes that could lead to more effective and targeted therapies for a variety of diseases. By reducing the time and cost associated with enzyme discovery, ProSSpeC represents a significant step forward in the field of protein engineering and precision medicine.