A systematic analysis of regression models for protein engineering

• Published in: PLoS computational biology 2024-05, Vol.20 (5), p.e1012061-e1012061
• Main Authors: Michael, Richard, Kæstel-Hansen, Jacob, Mørch Groth, Peter, Bartels, Simon, Salomon, Jesper, Tian, Pengfei, Hatzakis, Nikos S, Boomsma, Wouter
• Format: Article
• Language: English
• Subjects: Algorithms; Amino acids; Analysis; Biology and Life Sciences; Engineering; Engineering and Technology; Linear models (Statistics); Linear regression models; Machine learning; Methods; Mutation; Optimization; Physical Sciences; Protein engineering; Proteins; Regression analysis; Regression models; Research and Analysis Methods; Social Sciences
• ISSN: 1553-7358; 1553-734X; 1553-7358
• DOI: 10.1371/journal.pcbi.1012061

To optimize proteins for particular traits holds great promise for industrial and pharmaceutical purposes. Machine Learning is increasingly applied in this field to predict properties of proteins, thereby guiding the experimental optimization process. A natural question is: How much progress are we making with such predictions, and how important is the choice of regressor and representation? In this paper, we demonstrate that different assessment criteria for regressor performance can lead to dramatically different conclusions, depending on the choice of metric, and how one defines generalization. We highlight the fundamental issues of sample bias in typical regression scenarios and how this can lead to misleading conclusions about regressor performance. Finally, we make the case for the importance of calibrated uncertainty in this domain.