Computational enzyme metrics vary widely in predicting catalytic activity
Do in-silico metrics filter designed enzymes for catalysis?
Computational Engineering, Finance, and Science
Summary
Choosing which computer-designed enzymes to test in the lab is a hard problem because it's unclear how well computer scores predict how well enzymes will work. The authors looked at 33 different computer metrics to see if they could tell active enzymes from inactive ones, figure out what substrates enzymes prefer, rank enzyme variants by how active they are, or detect changes to important parts of the enzyme. They found that no single measure works well for all these tasks, and different metrics capture different aspects of enzyme function. This means enzyme designers should be careful when using these computer scores to decide which enzymes to experiment on.
What this means in practice
- •For enzyme engineering teams: Identify which computational metrics best separate active from inactive enzyme designs to prioritize lab testing of enzyme variants.
- •For biotech product developers: Use enzyme structure and activity metrics to enhance selection processes when developing enzymes for specific substrates or industrial processes.
Authors
Ziqi Zhang, Wei Zhang, Zhaohong Deng
Abstract
Computational metrics are widely used to prioritize enzyme designs for experimental testing, yet their ability to capture catalytic function remains poorly established. Here, we systematically evaluate 33 in-silico metrics across four complementary tasks: distinguishing active from inactive enzymes, identifying substrate specificity, ranking enzyme variants by activity, and detecting perturbations to catalytic residues. We assess these capabilities using natural enzymes, computationally designed enzymes, activity-altering variants, and controlled structural perturbations that compare catalytic residues with matched non-catalytic sites. No single metric consistently captures all aspects of catalytic function. In activity discrimination, catalytic-residue intra-residue repulsion achieves the highest AUROC among the evaluated metrics for both natural enzymes (0.61) and 192 designs that passed the filters of their original design campaign (0.80). For substrate discrimination, interface predicted TM-score (ipTM) performs best (AUROC 0.70), whereas the catalytic pKa range shows the strongest association with activity across single-point variants of BglB (Spearman's rho = 0.48). In controlled perturbation tests, the minimum per-chain predicted TM-score (pTM) has a mean probability of 0.75 of responding more strongly to perturbations at catalytic residues than to matched perturbations at non-catalytic sites. These results reveal substantial variation in what current computational metrics capture, underscoring the need to distinguish structural confidence and physical plausibility from direct evidence of catalytic function when prioritizing enzyme designs.