Supplemental files to the study "Limitations of Current Machine-Learning Models in Predicting Enzymatic Functions for Uncharacterized Proteins"

Thirty to seventy percent of proteins in any given genome have no assigned function and have been labeled as the protein “unknome”. This large knowledge shortfall is one of the final frontiers of biology. Machine-Learning (ML) approaches are enticing, with early successes demonstrating the ability to propagate functional knowledge from experimentally characterized proteins. An open question is the ability of machine-learning approaches to predict enzymatic functions unseen in the training sets. Using a set of E. coli unknowns, we evaluated the current state-of-the-art machine-learning approaches and found that these methods currently lack the ability to integrate scientific reasoning into their prediction algorithms. While human annotators are able to leverage the plethora of genomic data in making plausible predictions into the unknown, current ML methods not only fail to make novel predictions but also make basic logic errors in their predictions. This underscores the need to further develop ML methods and to deploy deterministic approaches to test for ‘hallucinations’ and other aberrant behavior in the current generation of predictive modeling. eXplainable AI (XAI) analysis revealed that noisy, ambiguous, or low contribution profiles across the protein sequence are strong indicators of unreliable predictions, enabling systematic identification of potential errors and elimination of most uncertain predictions. These findings demonstrate the value of integrating XAI-driven filtering strategies to improve the reliability of machine-learning-based protein annotation.