Machine learning enabling prediction of the bond dissociation enthalpy of hypervalent iodine from SMILES

Abstract

Machine learning to create models on the basis of big data enables predictions from new input data. Many tasks formerly performed by humans can now be achieved by machine learning algorithms in various fields, including scientific areas. Hypervalent iodine compounds (HVIs) have long been applied as useful reactive molecules. The bond dissociation enthalpy (BDE) value is an important indicator of reactivity and stability. Experimentally measuring the BDE value of HVIs is difficult, however, and the value has been estimated by quantum calculations, especially density functional theory (DFT) calculations. Although DFT calculations can access the BDE value with high accuracy, the process is highly time-consuming. Thus, we aimed to reduce the time for predicting the BDE by applying machine learning. We calculated the BDE of more than 1000 HVIs using DFT calculations, and performed machine learning. Converting SMILES strings to Avalon fingerprints and learning using a traditional Elastic Net made it possible to predict the BDE value with high accuracy. Furthermore, an applicability domain search revealed that the learning model could accurately predict the BDE even for uncovered inputs that were not completely included in the training data.

Comment on ‘Physics-based representations for machine learning properties of chemical reactions’

“…If we are interested in assessing model performance on new molecules, we can train a model with many reaction templates but use substructure splitting to create training, validation, and testing sets. Bemis-Murcko scaffolds [70] are commonly used to partition the data for this purpose, though clustering based on other input features or chemical similarity to measure extrapolation has also been explored [23,[71][72][73][74][75][76][77][78][79][80][81][82][83][84][85][86][87][88] as has quantifying domains of model applicability [89][90][91][92][93]. Scaffold splitting is not perfect, but by ensuring that molecules in the testing set are structurally different than those in the training set, it offers a much better assessment of generalizability than splitting randomly [17,24,67,[94][95][96][97][98][99][100][101][102][103][104][105][106][107][108][109]…”

Efficiently Predicting Reaction Rates and Revealing Reactive Sites with a Molecular Image-Vision Transformer and Fukui Function Validation

“…The static nature of convolutional kernel weights further restricts their adaptability to varying inputs. In contrast, the vision transformer (ViT), especially when pretrained on extensive data sets, demonstrates superior performance over traditional convolutional networks across several image recognition benchmarks, including ImageNet, CIFAR-100, and VTAB. Importantly, the integration of gradient-weighted class activation mapping (Grad-CAM) with density functional theory (DFT) in recent research highlights a methodological advancement.…”

“…This combination not only validates but also deepens our trust in the predictive capabilities of deep ViT models, bridging the gap between advanced image recognition techniques and chemical computational models. Recent developments in the field have highlighted the integration of machine learning with DFT calculations as a leading methodology for improving data set training and enhancing model prediction accuracies . Specifically, utilizing DFT outcomes in machine learning applications has facilitated swift and accurate forecasts of bond dissociation enthalpy values .…”

Developing Catalysts via Structure‐Property Relations Discovered by Machine Learning: An Industrial Perspective

“…One recent example was by Bello-Juardo et al [18] and Schwalbe-Koda et al [19] wherein they proposed a strategy to control the phase selectivity in templated zeolite synthesis by using ML-based strategy to make a catalyst for selective catalytic reduction (SCR). Major components of the remaining two-thirds of the set of deal with using ML for improving advance-characterization interpretation [20,21], accelerating quantum chemical simulations [22,23], or process simulations [24,25], which are not the focus of this perspective. Now, if compare the number of contributions from industry with academia, the numbers are even lower.…”