DiZyme: Open‐Access Expandable Resource for Quantitative Prediction of Nanozyme Catalytic Activity

Abstract

Enzymes suffer from high cost, complex purification, and low stability. Development of low‐cost artificial enzymes of comparative or higher effectiveness is desired. Given its complexity, it is desired to presume their activities prior to experiments. While computational approaches demonstrate success in modeling nanozyme activities, they require assumptions about the system to be made. Machine learning (ML) is an alternative approach towards data‐driven material property prediction achieving high performance even on multicomponent complex systems. Despite the growing demand for customized nanozymes, there is no open access nanozyme database. Here, a user‐friendly expandable database of >300 existing inorganic nanozymes is developed by data collection from >100 articles. Data analysis is performed to reveal the features responsible for catalytic activities of nanozymes, and new descriptors are proposed for its ML‐assisted prediction. A random forest regression (RFR) model for evaluation of nanozyme peroxidase activity is developed and optimized by correlation‐based feature selection and hyperparameter tuning, achieving performance up to R2 = 0.796 for Kcat and R2 = 0.627 for Km. Experiment‐confirmed unknown nanozyme activity prediction is also demonstrated. Moreover, the DiZyme expandable, open‐access resource containing the database, predictive algorithm, and visualization tool is developed to boost novel nanozyme discovery worldwide (https://dizyme.net).

Machine Learning Assisted Graphdiyne-Based Nanozyme Discovery

“…Razlivina et al established an open access database of nanozymes from over 100 published research papers and achieved high accuracy predictions of K m and K cat by the method of random forest. 27 Huang et al used deep neural network (DNN) algorithms to classify and quantitatively predict the enzyme-like activity exhibited by nanozymes and provided a promising strategy for developing nanozymes with desirable catalytic activity. 28 Despite their effectiveness in predicting nanozyme properties, the use of ML algorithms to guide nanozyme discovery is still fairly limited.…”

“…ML-driven approaches have been applied in numerous research domains successfully, such as catalysis, − chemistry, − materials, − and energy. , Recently, the ML technique has been used to predict the structure–activity relationship of nanozymes. Razlivina et al established an open access database of nanozymes from over 100 published research papers and achieved high accuracy predictions of K m and K cat by the method of random forest . Huang et al used deep neural network (DNN) algorithms to classify and quantitatively predict the enzyme-like activity exhibited by nanozymes and provided a promising strategy for developing nanozymes with desirable catalytic activity .…”

Machine Learning Guided Discovery of Superoxide Dismutase Nanozymes for Androgenetic Alopecia

“…Based on tens of thousands of materials information in the crystallographic databases and the materials project database, a SOD-like nanozyme data set was established covering 91 different M x P y S z ( x = 1–7, y = 1–4, z = 1–29), where the M x P y S z structures were consisted of the ordered unit cell and the transition metal atom coordinated with P y S z groups (Figure B). The feature selection is an important step in building ML algorithms because all predictors must be associated with features and ensured that they are predictable. , To make the model more effective, a correlation matrix algorithm was used to select descriptors with less correlation from the collected data set. Among them, Pearson’s correlation coefficient was utilized to determine the linear correlation or statistical relationship between two variables, and the threshold value for all descriptors pairs was 0.3.…”

“…The feature selection is an important step in building ML algorithms because all predictors must be associated with features and ensured that they are predictable. 33,34 To make the model more effective, a correlation matrix algorithm was used to select descriptors with less correlation from the collected data set. Among them, Pearson's correlation coefficient was utilized to determine the linear correlation or statistical relationship between two variables, and the threshold value for all descriptors pairs was 0.3.…”

Machine Learning Guided Graphdiyne based Nanozyme Discovery

“…The database provided general direction for the construction of nanozymes via a random forest regression model. 28 However, the precise prediction of any specific class of nanomaterial is unfeasible owing to the lack of sufficient data for accurate in-silico analysis. The varied performance descriptors and updates of learning models can significantly influence the prediction outcomes.…”