Machine Learning Models Based on Molecular Fingerprints and an Extreme Gradient Boosting Method Lead to the Discovery of JAK2 Inhibitors

Yang, Minjian; Tao, Bingzhong; Chen, Chengjuan; Jia, Wen‐Qiang; Sun, Shaolei; Zhang, Tiantai; Wang, Xiaojian

doi:10.1021/acs.jcim.9b00798

Cited by 42 publications

(31 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We want to emphasize that the substituent matching is just an ative way to evaluate the structural similarity;awide array of molecular features (i.e. molecular fingerprints, [43] molecular descriptors in RDKit, [20] physical organic descriptors, [16,17c,e] etc.) can be utilized to identify the related data for the designed hierarchical learning strategy based on the application purposes (vide infra).…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Towards Data‐Driven Design of Asymmetric Hydrogenation of Olefins: Database and Hierarchical Learning

Zhang

et al. 2021

Angewandte Chemie

View full text Add to dashboard Cite

Asymmetric hydrogenation of olefins is one of the most powerfula symmetric transformations in molecular synthesis.A lthough several privileged catalyst scaffolds are available,t he catalyst development for asymmetric hydrogenation is still atime-and resource-consuming process due to the lacko fp redictive catalyst design strategy.T argeting the data-driven design of asymmetric catalysis,weherein report the development of as tandardized database that contains the detailed information of over 12000 literature asymmetric hydrogenations of olefins.T his database provides av aluable platform for the machine learning applications in asymmetric catalysis.B ased on this database,w ed eveloped ah ierarchical learning approach to achieve predictive machine leaning model using only dozenso fe nantioselectivity data with the target olefin, which offers au seful solution for the few-shot learning problem and will facilitate the reaction optimization with new olefin substrate in catalysis screening.

show abstract

Section: Resultsmentioning

confidence: 99%

“…Through random splitting,90%of the related dataset awas used as training set, and the test set includes the other 10 %. A wide array of molecular descriptors based on 2D topological structure (i.e.R DKit [20] descriptors,M F, [43] etc. )o r3 D coordinate (i.e.A CSF, [44] MBTR, [45] etc.)…”

Section: Resultsmentioning

confidence: 99%

Towards Data‐Driven Design of Asymmetric Hydrogenation of Olefins: Database and Hierarchical Learning

Zhang

et al. 2021

Angewandte Chemie

View full text Add to dashboard Cite

show abstract

“…Rather than searching for similar molecules, machine learning models are trained to predict the activities of molecules based on their fingerprints. [8][9][10][11] This bypasses the need for similarity search but these approaches still rely, at its core, on precalculated fingerprints. A new class of ML algorithms, called Graph Neural Networks (GNN) are thought to overcome the calculation of fingerprints.…”

Section: Introductionmentioning

confidence: 99%

Using Domain-Specific Fingerprints Generated Through Neural Networks to Enhance Ligand-Based Virtual Screening

Menke

Koch

2021

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Molecular fingerprints are essential for different cheminformatics approaches like similarity-based virtual screening. In this work, the concept of neural (network) fingerprints in the context of similarity search is introduced in which the activation of the last hidden layer of a trained neural network represents the molecular fingerprint. The neural fingerprint performance of five different neural network architectures was analyzed and compared to the well-established Extended Connectivity Fingerprint (ECFP) and an autoencoder-based fingerprint. This is done using a published compound dataset with known bioactivity on 160 different kinase targets. We expect neural networks to combine information about the molecular space of already known bioactive compounds together with the information on the molecular structure of the query and by doing so enrich the fingerprint. The results show that indeed neural fingerprints can greatly improve the performance of similarity searches. Most importantly, it could be shown that the neural fingerprint performs well even for kinase targets that were not included in the training. Surprisingly, while Graph Neural Networks (GNNs) are thought to offer an advantageous alternative, the best performing neural fingerprints were based on traditional fully connected layers using the ECFP4 as input. The best performing kinase-specific neural fingerprint will be provided for public use.

show abstract

“…All docking poses were ranked according to their docking scores, in which lower score indicated lower binding energy. Structure-based docking methods rank molecules via a scoring function that is calculated based on their inner methods (Yang et al, 2019). In addition, the previous study also showed that AutoDock Vina was a more efficient option for stimulating protein docking with macrolides and analogues of intermediate size ligand compared to other docking programs, such as Glide 6.6, AutoDock 4.2 and DOCK 6.5 (Castro-Alvarez et al, 2017).…”

Section: Resultsmentioning

confidence: 99%

In silico study of anthocyanin and ternatin flavonoids for the treatment of inflammation-related diseases using molecular docking analysis

Wijaya¹,

Yulandi²,

Gunawan³

et al. 2020

Food Res.

View full text Add to dashboard Cite

Inflammatory markers such as cyclooxygenase (COX)-2, inducible nitric oxide synthase (iNOS), myeloperoxidase (MPO), and prostaglandin (PEG) are widely known as major targets in discovering natural anti-inflammatory drugs for the treatment of inflammationrelated diseases. Non-steroidal anti-inflammatory drugs (NSAIDs) such as ibuprofen and aspirin are mostly used at present, however, some NSAIDS have been reported to cause gastrointestinal side effect due to ligand-protein interaction. Molecular docking is a promising tool to study such modes of interaction. In this study, we evaluated the potential use of anthocyanin and ternatin flavonoids as natural anti-inflammatory agents for treatment of inflammatory-related diseases using in silico molecular docking assay. Automated docking study using Protein-Ligand ANT System (PLANTS) and AutoDock Vina was performed with various ligand molecules, including ibuprofen, anthocyanin, and ternatin against the protein crystal structures of COX-1, COX-2, iNOS, and MPO. The in silico data demonstrated that ibuprofen bound effectively to the active site of COX-1 and MPO with minimum binding energy, yet the compound required more energy to bind the active site of COX-2. Ternatin flavonoid was bound to COX-2 and iNOS with minimum binding energy. In terms of binding energy, anthocyanin flavonoid was found to be effective for inhibiting COX-1, COX-2, and iNOS. These results suggested that anthocyanin and ternatin flavonoids may potentially be developed as anti-inflammatory drug candidate for the treatment of inflammatory-related diseases.

show abstract

Machine Learning Models Based on Molecular Fingerprints and an Extreme Gradient Boosting Method Lead to the Discovery of JAK2 Inhibitors

Cited by 42 publications

References 52 publications

Towards Data‐Driven Design of Asymmetric Hydrogenation of Olefins: Database and Hierarchical Learning

Towards Data‐Driven Design of Asymmetric Hydrogenation of Olefins: Database and Hierarchical Learning

Using Domain-Specific Fingerprints Generated Through Neural Networks to Enhance Ligand-Based Virtual Screening

In silico study of anthocyanin and ternatin flavonoids for the treatment of inflammation-related diseases using molecular docking analysis

Contact Info

Product

Resources

About