Exploring QSAR models for activity-cliff prediction

Dablander, Markus; Hanser, Thierry; Lambiotte, Renaud; Morris, Garrett M.

doi:10.1186/s13321-023-00708-w

Cited by 16 publications

(15 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Among them, graph neural networks (GNNs) that utilize graph-based representations have gained signi cant popularity in predicting molecular properties 3,5,6 , as they enable a more direct and interpretable capture of the structural and relational information of molecules compared to traditional descriptors/ ngerprints-based ML models. However, contrary to common belief, recent studies have shown that traditional ML methods based on descriptors/ ngerprints outperform state-of-the-art end-to-end DL approaches in molecular property prediction [9][10][11] , particularly in the prediction of molecular activity with activity cliffs (ACs) for GNN models 10,11 . For example, van Tilborg et al (2022) 10 conducted a comprehensive study comparing the performance of graph-based DL models (GAT, GCN, AFP, and MPNN), SMILES string-based DL models (Transformer, CNN, and LSTM), and ngerprint-based ML models (MLP, KNN, GBM, RF, and SVM) on 30 activity benchmark datasets.…”

mentioning

confidence: 78%

Online triplet contrastive learning enables efficient cliff awareness in molecular activity prediction

Shen

Chen

Shi

et al. 2023

Preprint

View full text Add to dashboard Cite

Predicting molecular activity and quantitative structure-activity relationship (QSAR) is important for drug discovery and optimization. With molecular structures as frames, graph neural networks (GNNs) are suited for activity prediction but tend to overlook activity-cliffs (ACs) where structurally-similar molecules have vastly different activity values. To address this, we introduced a new online triplet contrastive learning framework ACANet that incorporates a unique activity-cliff-awareness (ACA) loss function, enabling efficient AC-awareness during training. The ACA loss enhances metric learning in the latent space and task learning in the target space simultaneously to make networks aware of ACs. ACANet outperformed the state-of-the-art machine learning and deep learning models in activity prediction and AC awareness on 39 benchmark datasets. ACA loss function is superior in AC-awareness than the mean absolute error and mean squared error loss functions. This innovative approach opens new avenues and provides valuable tools for applications in drug discovery and chemical engineering.

show abstract

mentioning

confidence: 78%

Online triplet contrastive learning enables efficient cliff awareness in molecular activity prediction

Shen

Chen

Shi

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…QSAR models, which assume that similar compounds have similar toxicity outcomes, often struggle to accurately predict activity cliffs (ACs) where this assumption does not hold. − ACs refer to pairs of small molecules that share high structural similarity but display a significant difference in their binding affinity toward a specific pharmacological target . Despite the difficulties posed by ACs, they provide a substantial understanding of the structure–activity relationship (SAR), which holds great value.…”

Section: Qsar and Its Limitationsmentioning

confidence: 99%

“…For example, AC prediction by QSAR models generally improved when one compound’s activity is known . Also, graph isomorphism features performed competitively for AC classification, while extended-connectivity fingerprints were best for general QSAR prediction . Traditional methods use a fixed potency difference, regardless of the target.…”

Section: Qsar and Its Limitationsmentioning

confidence: 99%

“…Nevertheless, the extraction and application of SAR information from AC clusters present considerable challenges and require systematic approaches to access SAR information . For example, AC prediction by QSAR models generally improved when one compound’s activity is known . Also, graph isomorphism features performed competitively for AC classification, while extended-connectivity fingerprints were best for general QSAR prediction .…”

Section: Qsar and Its Limitationsmentioning

confidence: 99%

See 1 more Smart Citation

A Review on the Recent Applications of Deep Learning in Predictive Drug Toxicological Studies

Sinha¹,

Ghosh²,

Sil

2023

Chem. Res. Toxicol.

View full text Add to dashboard Cite

Drug toxicity prediction is an important step in ensuring patient safety during drug design studies. While traditional preclinical studies have historically relied on animal models to evaluate toxicity, recent advances in deep-learning approaches have shown great promise in advancing drug safety science and reducing animal use in preclinical studies. However, deep-learning-based approaches also face challenges in handling large biological data sets, model interpretability, and regulatory acceptance. In this review, we provide an overview of recent developments in deep-learning-based approaches for predicting drug toxicity, highlighting their potential advantages over traditional methods and the need to address their limitations. Deep-learning models have demonstrated excellent performance in predicting toxicity outcomes from various data sources such as chemical structures, genomic data, and high-throughput screening assays. The potential of deep learning for automated feature engineering is also discussed. This review emphasizes the need to address ethical concerns related to the use of deep learning in drug toxicity studies, including the reduction of animal use and ensuring regulatory acceptance. Furthermore, emerging applications of deep learning in drug toxicity prediction, such as predicting drug–drug interactions and toxicity in rare subpopulations, are highlighted. The integration of deep-learning-based approaches with traditional methods is discussed as a way to develop more reliable and efficient predictive models for drug safety assessment, paving the way for safer and more effective drug discovery and development. Overall, this review highlights the critical role of deep learning in predictive toxicology and drug safety evaluation, emphasizing the need for continued research and development in this rapidly evolving field. By addressing the limitations of traditional methods, leveraging the potential of deep learning for automated feature engineering, and addressing ethical concerns, deep-learning-based approaches have the potential to revolutionize drug toxicity prediction and improve patient safety in drug discovery and development.

show abstract

“…Beginning in 2012, various attempts have been made to predict ACs. − Most of these studies have attempted to predict compound pairs forming ACs (often applying the MMP-cliff definition) and distinguish them from pairs of compounds with small potency differences. ,− For these purposes, ML classification models were used, often producing high accuracy in distinguishing ACs from other pairs of similar compounds. Recently, DNN variants have been used to predict ACs from molecular images , or graphs using representation learning. , By contrast, only few attempts have thus far been made to predict the actual potency value of AC compounds using ML regression models and/or DNNs of varying complexity. ,− The results of the currently most comprehensive study have confirmed the challenges in accurately predicting the potency of AC compounds and have shown that standard ML regression models yielded overall better performance than DNNs …”

Section: Introductionmentioning

confidence: 99%

Anatomy of Potency Predictions Focusing on Structural Analogues with Increasing Potency Differences Including Activity Cliffs

Janela,

Bajorath

2023

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Potency predictions are popular in compound design and optimization but are complicated by intrinsic limitations. Moreover, even for nonlinear methods, activity cliffs (ACs, formed by structural analogues with large potency differences) represent challenging test cases for compound potency predictions. We have devised a new test system for potency predictions, including AC compounds, that is based on partitioned matched molecular pairs (MMP) and makes it possible to monitor prediction accuracy at the level of analogue pairs with increasing potency differences. The results of systematic predictions using different machine learning and control methods on MMP-based data sets revealed increasing prediction errors when potency differences between corresponding training and test compounds increased, including large prediction errors for AC compounds. At the global level, these prediction errors were not apparent due to the statistical dominance of analogue pairs with small potency differences. Test compounds from such pairs were accurately predicted and determined the observed global prediction accuracy. Shapley value analysis, an explainable artificial intelligence approach, was applied to identify structural features determining potency predictions using different methods. The analysis revealed that numerical predictions of different regression models were determined by features that were shared by MMP partner compounds or absent in these compounds, with opposing effects. These findings provided another rationale for accurate predictions of similar potency values for structural analogues and failures in predicting the potency of AC compounds.

show abstract

Exploring QSAR models for activity-cliff prediction

Cited by 16 publications

References 48 publications

Online triplet contrastive learning enables efficient cliff awareness in molecular activity prediction

Online triplet contrastive learning enables efficient cliff awareness in molecular activity prediction

A Review on the Recent Applications of Deep Learning in Predictive Drug Toxicological Studies

Anatomy of Potency Predictions Focusing on Structural Analogues with Increasing Potency Differences Including Activity Cliffs

Contact Info

Product

Resources

About