Out-of-the-box deep learning prediction of pharmaceutical properties by broadly learned knowledge-based molecular representations

Shen, Wanxiang; Zeng, Xian; Zhu, Feng; Wang, Ya li; Chu, Qing; Tan, Ying; Jiang, Yu; Chen, Yu Zong

doi:10.1038/s42256-021-00301-6

Cited by 90 publications

(76 citation statements)

References 58 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The dataset was split into training, validation, and the test set follow the ratio of 8/1/1. Table 1 reports the best achieved AUC-ROC of our method is 0.763, which is better than results reported in Chemprop 46 (AUC-ROC = 0.738), MMNB 47 (AUC-ROC = 0.739), and MoleculeNet 9 (AUC-ROC ECFP = 0.671).…”

Section: Algebraic Graph-assisted Deep Bidirectional Transformer (Agbt)mentioning

confidence: 64%

“…And the final score for the dataset is the average score over ten different data-splittings. As shown in Table 1, using the STDNN algorithm, our method obtains the best RMSE of 0.994 on FreeSolv, which is better than the best results reported by Chemprop 46 (RMSE = 1.075), MMNB 47 (RMSE = 1.155), and MoleculeNet 9 (RMSE GraphConv = 1.15). For the Lipophilicity dataset, using the RF algorithm, the RF method, our descriptors obtains the best RMSE of 0.573 ± 0.026, a result that is close to the best result reported by Chemprop 46 (RMSE GraphCov = 0.555).…”

Section: Algebraic Graph-assisted Deep Bidirectional Transformer (Agbt)mentioning

confidence: 72%

See 1 more Smart Citation

Algebraic graph-assisted bidirectional transformers for molecular property prediction

et al. 2021

View full text Add to dashboard Cite

The ability of molecular property prediction is of great significance to drug discovery, human health, and environmental protection. Despite considerable efforts, quantitative prediction of various molecular properties remains a challenge. Although some machine learning models, such as bidirectional encoder from transformer, can incorporate massive unlabeled molecular data into molecular representations via a self-supervised learning strategy, it neglects three-dimensional (3D) stereochemical information. Algebraic graph, specifically, element-specific multiscale weighted colored algebraic graph, embeds complementary 3D molecular information into graph invariants. We propose an algebraic graph-assisted bidirectional transformer (AGBT) framework by fusing representations generated by algebraic graph and bidirectional transformer, as well as a variety of machine learning algorithms, including decision trees, multitask learning, and deep neural networks. We validate the proposed AGBT framework on eight molecular datasets, involving quantitative toxicity, physical chemistry, and physiology datasets. Extensive numerical experiments have shown that AGBT is a state-of-the-art framework for molecular property prediction.

show abstract

Section: Algebraic Graph-assisted Deep Bidirectional Transformer (Agbt)mentioning

confidence: 64%

Section: Algebraic Graph-assisted Deep Bidirectional Transformer (Agbt)mentioning

confidence: 72%

Algebraic graph-assisted bidirectional transformers for molecular property prediction

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Such comparative data include the knowledge of poor binders or non-binders of individual target that are useful for developing drug discovery tool of enhanced performance ( 5–7 ); the information of prodrugs that facilitates drug design by improving pharmacokinetic/pharmacodynamic features ( 8 ); the co-targets of therapeutic targets that facilitate the investigations of multi-target strategies ( 9 ), off-target ( 10 , 11 ) & undesired effect ( 9 ); the collective structure-activity landscapes of drugs against individual target that reveal important pharmaceutical features such as activity cliffs ( 12 ); and the drugs’ profiles of their drug-like properties that provide drug-likeness landscapes of the explored bioactive chemical space for therapeutic targets ( 13 ). Particularly, there is a rapid trend of the discovery of Artificial Intelligence (AI) tools for the drug discovery ( 14 , 15 ), including the AI tools for identifying bioactive compounds, and the construction of such tools requires data of poor binders and non-binders of a specific target ( 16 ). In the meantime, the existing prodrug data may inspire new ideas to avoid the drug development challenges that limit formulation option or result in undesired biopharmaceutical/pharmacokinetic performance ( 8 ).…”

Section: Introductionmentioning

confidence: 99%

Therapeutic target database update 2022: facilitating drug discovery with enriched comparative data of targeted agents

Zhou

Zhang

Lian

et al. 2021

Nucleic Acids Research

Self Cite

411

241

View full text Add to dashboard Cite

Drug discovery relies on the knowledge of not only drugs and targets, but also the comparative agents and targets. These include poor binders and non-binders for developing discovery tools, prodrugs for improved therapeutics, co-targets of therapeutic targets for multi-target strategies and off-target investigations, and the collective structure-activity and drug-likeness landscapes of enhanced drug feature. However, such valuable data are inadequately covered by the available databases. In this study, a major update of the Therapeutic Target Database, previously featured in NAR, was therefore introduced. This update includes (a) 34 861 poor binders and 12 683 non-binders of 1308 targets; (b) 534 prodrug-drug pairs for 121 targets; (c) 1127 co-targets of 672 targets regulated by 642 approved and 624 clinical trial drugs; (d) the collective structure-activity landscapes of 427 262 active agents of 1565 targets; (e) the profiles of drug-like properties of 33 598 agents of 1102 targets. Moreover, a variety of additional data and function are provided, which include the cross-links to the target structure in PDB and AlphaFold, 159 and 1658 newly emerged targets and drugs, and the advanced search function for multi-entry target sequences or drug structures. The database is accessible without login requirement at: https://idrblab.org/ttd/.

show abstract

“…Graph learning methods have the potential to improve drug discovery efficiency dramatically for their ability to amplify insights available from existing drug-related datasets 5 . Using the insights to predict molecular interactions and properties 6,7 is a key to finding potential drug candidates from the vast chemical space with extremely fast speed and low cost. On the other hand, molecular generation 8,9 based on the insights can more efficiently traverse the vast chemical space to find potential drug candidates.…”

Section: Introductionmentioning

confidence: 99%

“…More recently, a few works have been reported to address these problems above. MolMapNet introduced an out-of-the-box deep learning method based on broadly learning knowledge-based representations to get reliable prediction performance on more datasets 7 . A recent work introduced a neural architecture search based method to design neural architecture for molecular property predictions 11 .…”

Section: Introductionmentioning

confidence: 99%

GLAM: An adaptive graph learning method for automated molecular interactions and properties predictions

Hsieh

et al. 2022

Preprint

View full text Add to dashboard Cite

Improving drug discovery efficiency is a core and long-standing challenge in drug discovery. For this purpose, many graph learning methods have been developed to search potential drug candidates with fast speed and low cost. In fact, the pursuit of high prediction performance on a limited number of datasets has crystallized them, making them lose advantage in repurposing to new data generated in drug discovery. Here we propose a flexible method that can adapt to any dataset and make accurate predictions. The proposed method employs an adaptive pipeline to learn from a dataset and output a predictor. Without any manual intervention, the method achieves far better prediction performance on all tested datasets than traditional methods, which are based on hand-designed neural architectures and other fixed items. In addition, we found that the proposed method is more robust than traditional methods and can provide meaningful interpretability. Given the above, the proposed method can serve as a reliable method to predict molecular interactions and properties with high adaptability, performance, robustness and interpretability. This work would take a solid step forward to the purpose of aiding researchers to design better drugs with high efficiency.

show abstract

Out-of-the-box deep learning prediction of pharmaceutical properties by broadly learned knowledge-based molecular representations

Cited by 90 publications

References 58 publications

Algebraic graph-assisted bidirectional transformers for molecular property prediction

Algebraic graph-assisted bidirectional transformers for molecular property prediction

Therapeutic target database update 2022: facilitating drug discovery with enriched comparative data of targeted agents

GLAM: An adaptive graph learning method for automated molecular interactions and properties predictions

Contact Info

Product

Resources

About