Co‐crystal Prediction by Artificial Neural Networks**

Devogelaer, Jan-Joris; Meekes, Hugo; Tinnemans, Paul; Vlieg, Elias; Gelder, R. de

doi:10.1002/ange.202009467

Cited by 9 publications

(22 citation statements)

References 70 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Machine learning-based models with high accuracy are typically trained by experimentally constructing an in-house negative data set for a certain type of compound, and thus have a narrow applicability domain. By constructing comprehensive negative samples via a randomly chosen approach or other computational methods, recent studies have begun to overcome this drawback . The generated negative samples would inevitably contain potential cocrystals that have not been observed, which might affect the accuracy of the predictive models.…”

Section: Discussionmentioning

confidence: 99%

Strategy for Efficient Discovery of Cocrystals via a Network-Based Recommendation Model

Zheng

Zhu

et al. 2020

Crystal Growth & Design

View full text Add to dashboard Cite

Experimental screening of cocrystals is usually laborious and time-consuming; therefore, it is urgent to develop effective in silico predictive models to guide cocrystal discovery. In this study, network-based recommendation models were proposed to predict new cocrystals for molecules in cocrystal network. The local random walk (LRW) recommender algorithm was first confirmed as an effective model in cocrystal design. The algorithmic principle of LRW could capture the supramolecular synthon mechanisms in the cocrystal system and grasp the structural features of the cocrystal network, thus possessing satisfactory predictive capability. Various pharmaceutical cocrystals reported in the recent literature could be distinguished by our model, which demonstrates the good generalization capability inherent in our approach. As a case study, new cocrystals for apatinib were predicted and subsequently obtained. The consistency between prediction and experimental results highlighted the accuracy and practicability of the predictive model. Particularly, our predictive model is competitive in computational time and easy to implement. In summary, our network-based recommendation model would be an effective tool to guide experimental cocrystal screening and improve the efficiency of cocrystal discovery.

show abstract

Section: Discussionmentioning

confidence: 99%

Strategy for Efficient Discovery of Cocrystals via a Network-Based Recommendation Model

Zheng

Zhu

et al. 2020

Crystal Growth & Design

View full text Add to dashboard Cite

show abstract

“…Graph convolutional neural networks do this by using graph convolutional layers, which can process node information passing through edges by combining linear and nonlinear transformations and pooling. , Finally, the corresponding features of the molecules will be extracted by the readout layer and generated as a one-dimensional sequence, which is further processed by the neural network to give the final output. In this research, the GCNs is built based on the model structure and hyperparameters reported by Devogelaer in 2020 . Five GCN were randomly trained on training datasets and further ensembled together to finally judge whether a cocrystal is formed.…”

Section: Machine Learning Modelingmentioning

confidence: 99%

“…Devogelaer and his colleagues also collected a positive dataset from CSD and constructed a dummy negative dataset using their own algorithm. They then ensembled a graph convolutional network (GCN) based on atomic features and an ECFP-based neural network (FPNN) as the cocrystal screening model and finally obtained about 80% accuracy and one new cocrystal …”

Section: Introductionmentioning

confidence: 99%

Investigating Spatial Charge Descriptors for Prediction of Cocrystal Formation Using Machine Learning Algorithms

Hao

Hung

Shimoyama

2022

Crystal Growth & Design

View full text Add to dashboard Cite

Recently, drug modification via cocrystals has attracted great attention due to its high flexibility for the modulation of drug physicochemical properties. To reduce the cost of screening experiments, machine learning (ML) algorithms have proven to be one of the most effective ways to rapidly screen cocrystal formation. However, the choice of molecular descriptors has a significant impact on its prediction accuracy. In this work, two space-charge descriptors (COSMO-based σ-profile and three-dimensional (3D) spatial charge) are introduced into ML modeling, and two innovative COSMO-SVM and 3D-CNN ML algorithms are developed to screen the cocrystal formation. The two proposed ML models using space-charge descriptors demonstrate superior predictive accuracy for cocrystal formation compared to preproposed ML methods using extended connectivity fingerprints (ECFPs) and graph convolutional networks (GCNs). Considering computational efficiency and accuracy, 3D-CNN would be a very efficient ML model in the cocrystal formation screening task. In addition, the formation of two new cocrystals of norfloxacin was predicted by machine learning, and the cocrystals were obtained experimentally.

show abstract

“…The method can be used as a preliminary screening tool for excluding the possible formation of immiscible solid states. Devogelaer et al 35 introduced two neural network models that accept a pair of molecules as input and classify whether they can form a cocrystal. Two models differed in their input molecular representations and initial preprocessing steps.…”

Section: Ai and Cocrystalmentioning

confidence: 99%

Progress in Research on Artificial Intelligence Applied to Polymorphism and Cocrystal Prediction

et al. 2021

View full text Add to dashboard Cite

Artificial intelligence (AI) is a technology that builds an artificial system with certain intelligence and uses computer software and hardware to simulate intelligent human behavior. When combined with drug research and development, AI can considerably shorten this cycle, improve research efficiency, and minimize costs. The use of machine learning to discover novel materials and predict material properties has become a new research direction. On the basis of the current status of worldwide research on the combination of AI and crystal form and cocrystal, this mini-review analyzes and explores the application of AI in polymorphism prediction, crystal structure analysis, crystal property prediction, cocrystal former (CCF) screening, cocrystal composition prediction, and cocrystal formation prediction. This study provides insights into the future applications of AI in related fields.

show abstract

Co‐crystal Prediction by Artificial Neural Networks**

Cited by 9 publications

References 70 publications

Strategy for Efficient Discovery of Cocrystals via a Network-Based Recommendation Model

Strategy for Efficient Discovery of Cocrystals via a Network-Based Recommendation Model

Investigating Spatial Charge Descriptors for Prediction of Cocrystal Formation Using Machine Learning Algorithms

Progress in Research on Artificial Intelligence Applied to Polymorphism and Cocrystal Prediction

Contact Info

Product

Resources

About