Abstract:A machine-learning model trained on the whole Cambridge Structural Database was developed to assist high-throughput cocrystal screening. With only 2D structures taken as inputs, the probability of cocrystal formation is returned for two given molecules. All of the cocrystal records in the CSD were used as positive samples, while negative samples were constructed by randomly combining different molecules into chemical pairs. Our model showed a prediction ability comparable with that of a widely used ab initio method in a head-to-head comparison test. Both experimental and virtual cocrystal screening against captopril were conducted at the same time to further validate the model. Two cocrystals of captopril with l-proline and sarcosine were obtained and characterized by PXRD, DSC, and FT-IR. These two coformers were also successfully predicted by our model. These results suggest that the tool we developed can be used to effectively guide coformer selection in the discovery of new cocrystals.The Supporting Information is available free of charge at <a class="ext-link" href="/doi/10.1021/acs.cgd.0c00767?goto=supporting-info">https://pubs.acs.org/doi/10.1021/acs.cgd.0c00767</a>.Detailed explanation of the chemical pair encoding method, ROC curves of different ML models using eight fingerprints (except MACCS and PubChem) on the test set, cocrystal probability ranking by the MACCS-based RF and PubChem-based RF for the 54 coformers against captopril, 1H NMR of CapPro and CapSar, hyperparameter searching results for ML models, scoring list of the consensus model on each API in head-to-head comparison tests with the MEPS-based method, brief explanation for symbols used in <a class="internalNav" href="#fig4">Figure </a><a class="internalNav" href="#fig4">4</a>, counts of values of top 20 important fingerprint bits on the whole data set, DSC data of Cap and cocrystals, and pKa1 values of Cap, Pro, and Sar (<a class="ext-link" href="/doi/suppl/10.1021/acs.cgd.0c00767/suppl_file/cg0c00767_si_001.pdf">PDF</a>)This article has not yet been cited by other publications.

Cocrystal Prediction Tool (CCPT): A Web Server for Deep Learning-Assisted Cocrystal Screening and Density Evaluation

Cocrystal virtual screening based on the XGBoost machine learning model

TCSP: a Template based crystal structure prediction algorithm and web server for materials discovery

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

Pharmaceutical Cocrystal Discovery via 3D-SMINBR: A New Network Recommendation Tool Augmented by 3D Molecular Conformations

Deep Supramolecular Language Processing for Co-crystal Prediction

A Robust Crystal Structure Prediction Method to Support Small Molecule Drug Development with Large Scale Validation and Prospective Studies

General Graph Neural Network-Based Model To Accurately Predict Cocrystal Density and Insight from Data Quality and Feature Representation

Machine Learning-Guided Prediction of Cocrystals Using Point Cloud-Based Molecular Representation

SMINBR: an Integrated Network and Chemoinformatics Tool Specialized for Prediction of Two-Component Crystal Formation.

TCSP: a Template-Based Crystal Structure Prediction Algorithm for Materials Discovery

Leveraging Deep Chemical Language Processing for Co-crystal Prediction

Machine-Learning-Guided Cocrystal Prediction Based on Large Data Base

AlphaCrystal-II: Distance matrix based crystal structure prediction using deep learning

End-to-End Crystal Structure Prediction from Powder X-Ray Diffraction

Enhancing crystal structure prediction by combining computational and experimental data via graph networks

Coupling complementary strategy to flexible graph neural network for quick discovery of coformer in diverse co-crystal materials

AlphaCrystal: Contact map based crystal structure prediction using deep learning

Reliable and Practical Computational Prediction of Molecular Crystal Polymorphs

DeepChargePredictor : a web server for predicting QM-based atomic charges via state-of-the-art machine-learning algorithms

D3CARP: a comprehensive platform with multiple-conformation based docking, ligand similarity search and deep learning approaches for target prediction and virtual screening