Predict protein location
I was fortunate to contribute to the work of very talented scientists at Moderna who developed a novel approach to predict bacterial protein topology.
This study introduces mtx-COBRA, a novel bioinformatics pipeline designed to predict the subcellular localization (SCL) of bacterial proteins—a crucial step in vaccine development. While bacterial proteins serve as potential antigens for immune system recognition, traditional tools like PSORTb often classify many proteins with an “Unknown” SCL, limiting vaccine target identification. mtx-COBRA overcomes this challenge by integrating Meta’s Evolutionary Scale Modeling with an Extreme Gradient Boosting machine learning model, enabling more accurate SCL predictions based solely on amino acid sequences. Trained on a curated dataset from UniProt and ePSORTdb, this tool enhances the efficiency of antigen discovery, advancing infectious disease research and vaccine design.
This work was published in Computers in Biology and Medicine in 2024.
Follow the link below to access the paper: