Learning Drug Function from Chemical Structure with Convolutional Neural Networks and Random Forests

Meyer, Jesse G.; Liu, Shengchao; Miller, Ian; Coon, Joshua J.; Gitter, Anthony

doi:10.1101/482877

“…A naive approach that utilizes such graphical information is to consider pictures of molecular graphs as inputs to a DNN. This has been tried in the literature [ 93 ] using 2D CNN architectures that have been effective for image classification [ 93 , 121 ]. However, these models do not outperform MLPs and random forests trained on ECFPs.…”

Section: Molecular Feature Representationsmentioning

confidence: 99%

Machine learning and AI-based approaches for bioactive ligand discovery and GPCR-ligand recognition

Raschka

¹

,

Kaufman

²

2020

Methods

View full text Add to dashboard Cite

In the last decade, machine learning and artificial intelligence applications have received a significant boost in performance and attention in both academic research and industry. The success behind most of the recent state-of-the-art methods can be attributed to the latest developments in deep learning. When applied to various scientific domains that are concerned with the processing of non-tabular data, for example, image or text, deep learning has been shown to outperform not only conventional machine learning but also highly specialized tools developed by domain experts. This review aims to summarize AI-based research for GPCR bioactive ligand discovery with a particular focus on the most recent achievements and research trends. To make this article accessible to a broad audience of computational scientists, we provide instructive explanations of the underlying methodology, including overviews of the most commonly used deep learning architectures and feature representations of molecular data. We highlight the latest AI-based research that has led to the successful discovery of GPCR bioactive ligands. However, an equal focus of this review is on the discussion of machine learning-based technology that has been applied to ligand discovery in general and has the potential to pave the way for successful GPCR bioactive ligand discovery in the future. This review concludes with a brief outlook highlighting the recent research trends in deep learning, such as active learning and semi-supervised learning, which have great potential for advancing bioactive ligand discovery.

show abstract

“…A naive approach that utilizes such graphical information is to consider pictures of molecular graphs as inputs to a DNN. This has been tried in the literature using 2D CNN architectures that have been effective for image classification [127,128]. However, these models do not outperform MLP and random forest trained on ECFPs.…”

Section: Graph Representationmentioning

confidence: 99%

Machine learning and AI-based approaches for bioactive ligand discovery and GPCR-ligand recognition

Raschka,

Kaufman

2020

Preprint

View full text Add to dashboard Cite

In the last decade, machine learning and artificial intelligence applications have received a significant boost in performance and attention in both academic research and industry. The success behind most of the recent state-of-the-art methods can be attributed to the latest developments in deep learning. When applied to various scientific domains that are concerned with the processing of non-tabular data, for example, image or text, deep learning has been shown to outperform not only conventional machine learning but also highly specialized tools developed by domain experts. This review aims to summarize AI-based research for GPCR bioactive ligand discovery with a particular focus on the most recent achievements and research trends. To make this article accessible to a broad audience of computational scientists, we provide instructive explanations of the underlying methodology, including overviews of the most commonly used deep learning architectures and feature representations of molecular data. We highlight the latest AI-based research that has led to the successful discovery of GPCR bioactive ligands. However, an equal focus of this review is on the discussion of machine learning-based technology that has been applied to ligand discovery in general and has the potential to pave the way for successful GPCR bioactive ligand discovery in the future. This review concludes with a brief outlook highlighting the recent research trends in deep learning, such as active learning and semi-supervised learning, which have great potential for advancing bioactive ligand discovery. AbbreviationsAI Artificial intelligence CNN Convolutional neural network DNN Deep neural network ECFP Extended connectivity fingerprint GNN Graph neural network GPCR G protein-coupled receptor GPU Graphics processing unit GRU Gated recurrent unit MLP Multilayer perceptron NN Neural network QSAR Quantitative structure-activity relationship RNN Recurrent neural network

show abstract

Druggability of Pharmaceutical Compounds Using Lipinski Rules with Machine Learning

Nhlapho,

Nyathi,

Ngwenya

et al. 2024

sciphar

0

View full text Add to dashboard Cite

In the field of pharmaceutical research, identifying promising pharmaceutical compounds is a critical challenge. The observance of Lipinski's Rule of Five (RO5) is a fundamental criterion, but evaluating many compounds manually requires significant resources and time. However, the integration of computational techniques in drug discovery in its early stages has significantly transformed the pharmaceutical industry, enabling further efficient screening and selection of possible drug candidates. Therefore, this study explores RO5 using algorithms of Machine Learning (ML), offering a comprehensive method to predict the druggability of pharmaceutical compounds. The study developed, evaluated, and validated the performance metrics of multiple supervised machine learning models. The best model was used to build an application that can predict and classify potential drug candidates. The findings revealed promising capabilities across all models for drug classification. Among all the explored models, Random Forest (RF), Extreme Gradient Boost (XGBoost), and Decision Tree (DT) classifiers demonstrated exceptional performance, achieving near-perfect accuracy of 99.94%, 99.81% and 99.87% respectively. This highlights the robustness of ensemble learning methods in classifying compounds based on RO5 adherence. The comparative analysis of these models underscores the importance of considering balanced accuracy, precision, F1-score, recall, and Receiver Operating Characteristics-Area Under the Curve (ROC-AUC) score, interpretability, and computational efficiency when choosing between ML algorithms in drug discovery. The DrugCheckMaster application was subsequently developed using the most predictive model and is now available on Render (https://capstone-project-dc7w.onrender.com/).

show abstract

Learning Drug Function from Chemical Structure with Convolutional Neural Networks and Random Forests

Cited by 3 publications

References 47 publications

Machine learning and AI-based approaches for bioactive ligand discovery and GPCR-ligand recognition

Machine learning and AI-based approaches for bioactive ligand discovery and GPCR-ligand recognition

Machine learning and AI-based approaches for bioactive ligand discovery and GPCR-ligand recognition

Druggability of Pharmaceutical Compounds Using Lipinski Rules with Machine Learning

Contact Info

Product

Resources

About