Machine learning‐based prediction of enzyme substrate scope: Application to bacterial nitrilases

Abstract

Predicting the range of substrates accepted by an enzyme from its amino acid sequence is challenging. Although sequence‐ and structure‐based annotation approaches are often accurate for predicting broad categories of substrate specificity, they generally cannot predict which specific molecules will be accepted as substrates for a given enzyme, particularly within a class of closely related molecules. Combining targeted experimental activity data with structural modeling, ligand docking, and physicochemical properties of proteins and ligands with various machine learning models provides complementary information that can lead to accurate predictions of substrate scope for related enzymes. Here we describe such an approach that can predict the substrate scope of bacterial nitrilases, which catalyze the hydrolysis of nitrile compounds to the corresponding carboxylic acids and ammonia. Each of the four machine learning models (logistic regression, random forest, gradient‐boosted decision trees, and support vector machines) performed similarly (average ROC = 0.9, average accuracy = ~82%) for predicting substrate scope for this dataset, although random forest offers some advantages. This approach is intended to be highly modular with respect to physicochemical property calculations and software used for structural modeling and docking.

The substrate scopes of enzymes: a general prediction model based on machine and deep learning

“…We compared ESP with two recently published models for predicting the substrate scope of specific enzyme families. Mou et al 3 We added all training data from Ref. 3 to our training set and validated the updated ESP model on the corresponding test data, which had no overlap with our training data.…”

“…ESP can be applied successfully across widely different enzymes and a broad range of metabolites. It outperforms recently published models designed for individual, well-studied enzyme families, which use much more detailed input data [3,4].…”

“…Existing machine learning approaches for predicting enzyme-substrate pairs were developed specifically for small enzyme families for which unusually comprehensive training datasets are available 3,4,16,17,18 .…”

“…For example, Mou et al 3 developed models to predict the substrates of bacterial nitrilases, using input features based on the 3D-structures and active sites of the enzymes. They trained various machine learning models based on experimental evidence for all possible enzyme-small molecule combinations within the models' prediction scope (N = 240).…”

New Trends in Bioremediation Technologies Toward Environment-Friendly Society: A Mini-Review

“…However, engineering enzymes in the data-assisted synthetic biology landscape could accelerate the hunt of the “super-enzyme” in environmental perspectives. However, as this is a new frontier to the scientific literature body, only a handful of the kindest efforts are available at present (Ajjolli Nagaraja et al, 2020 ; Lawson et al, 2020 ; Mou et al, 2020 ; Robinson et al, 2020 ; Siedhoff et al, 2020 ; Wittmann et al, 2020 ). Herein, we have summarized current state-of-the-art knowledge of the data-assisted enzyme redesigning ( Figure 1 ) to promote new studies on enzyme redesigning from an environmental perspective.…”