{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"bioRxiv"}],"indexed":{"date-parts":[[2026,9,18]],"date-time":"2026-09-18T07:04:01Z","timestamp":1789715041498,"version":"4.0.1"},"posted":{"date-parts":[[2026,9,14]]},"group-title":"Bioinformatics","reference-count":22,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T00:00:00Z","timestamp":1789344000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"award":["R35GM150620"],"award-info":[{"award-number":["R35GM150620"]}],"id":[{"id":"https:\/\/ror.org\/01cwqze88","id-type":"ROR","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2026,9,14]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                <jats:p>Activity cliffs, defined as structurally similar molecules with vastly different properties represent a fundamental challenge in modern day drug discovery for property prediction models. While Graph Neural Networks (GNNs) have advanced molecular property prediction, they inherently struggle with this problem due to representation collapse and node over smoothening. In this work, we evaluate the efficacy of incorporating various contrastive learning-based loss functions, including the Supervised Contrastive (SupCon) configurations into several GNN backbones to navigate this structure-activity landscape. Tested across the MoleculeACE benchmarks, the results indicate that while contrastive losses prevent GNN mode collapse to some extent and drastically improve the property-based separation in the embedding space, their impact on absolute predictive accuracy remains highly dataset dependent. Ultimately, these methods offer critical gains in interpretability for mapping optimized lead generations, opening up new directions for models that respect these sharp pharmacological discontinuities.<\/jats:p>","DOI":"10.64898\/2026.09.08.750257","type":"posted-content","created":{"date-parts":[[2026,9,15]],"date-time":"2026-09-15T04:15:22Z","timestamp":1789445722000},"source":"Crossref","is-referenced-by-count":0,"title":["How effective are contrastive learning-based approaches for activity-cliff prediction?"],"prefix":"10.64898","author":[{"ORCID":"https:\/\/orcid.org\/0009-0007-7910-6535","authenticated-orcid":false,"given":"Akash","family":"Surendran","sequence":"first","affiliation":[{"name":"Quantum Theory Project and Department of Chemistry, University of Florida, Gainesville, Florida 32611, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2121-4449","authenticated-orcid":false,"given":"Ram\u00f3n Alain","family":"Miranda-Quintana","sequence":"additional","affiliation":[{"name":"Quantum Theory Project and Department of Chemistry, University of Florida, Gainesville, Florida 32611, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"54368","reference":[{"key":"2026091723400788000_2026.09.08.750257v1.1","doi-asserted-by":"publisher","DOI":"10.1021\/jm201706b"},{"key":"2026091723400788000_2026.09.08.750257v1.2","doi-asserted-by":"publisher","DOI":"10.1016\/j.drudis.2014.02.003"},{"key":"2026091723400788000_2026.09.08.750257v1.3","doi-asserted-by":"crossref","first-page":"14360","DOI":"10.1021\/acsomega.9b02221","article-title":"Evolving concept of activity cliffs","volume":"4","year":"2019","journal-title":"ACS omega"},{"key":"2026091723400788000_2026.09.08.750257v1.4","unstructured":"Johnson, M. A. ; Maggiora, G. M. , et al. Concepts and applications of molecular similarity. (No Title) 1990."},{"key":"2026091723400788000_2026.09.08.750257v1.5","doi-asserted-by":"crossref","first-page":"100077","DOI":"10.1016\/j.aichem.2024.100077","article-title":"Molecular similarity: Theory, applications, and perspectives","volume":"2","year":"2024","journal-title":"Artificial Intelligence Chemistry"},{"key":"2026091723400788000_2026.09.08.750257v1.6","doi-asserted-by":"publisher","DOI":"10.1021\/ci9800211"},{"key":"2026091723400788000_2026.09.08.750257v1.7","doi-asserted-by":"crossref","unstructured":"Maggiora, G. M. On outliers and activity cliffs why QSAR often disappoints, 2006.","DOI":"10.1021\/ci060117s"},{"key":"2026091723400788000_2026.09.08.750257v1.8","doi-asserted-by":"publisher","DOI":"10.1021\/jm401120g"},{"key":"2026091723400788000_2026.09.08.750257v1.9","doi-asserted-by":"crossref","first-page":"181","DOI":"10.1002\/minf.201600023","article-title":"Advances in activity cliff research","volume":"35","year":"2016","journal-title":"Molecular informatics"},{"key":"2026091723400788000_2026.09.08.750257v1.10","unstructured":"Gilmer, J. ; Schoenholz, S. S. ; Riley, P. F. ; Vinyals, O. ; Dahl, G. E. In International conference on machine learning, 2017, pp 1263\u20131272."},{"key":"2026091723400788000_2026.09.08.750257v1.11","doi-asserted-by":"publisher","DOI":"10.1021\/acs.jcim.3c01250"},{"key":"2026091723400788000_2026.09.08.750257v1.12","doi-asserted-by":"publisher","DOI":"10.1021\/acs.jcim.9b00237"},{"key":"2026091723400788000_2026.09.08.750257v1.13","doi-asserted-by":"publisher","DOI":"10.1021\/acs.jcim.2c01073"},{"key":"2026091723400788000_2026.09.08.750257v1.14","doi-asserted-by":"crossref","first-page":"47","DOI":"10.1186\/s13321-023-00708-w","article-title":"Exploring QSAR models for activity-cliff prediction","volume":"15","year":"2023","journal-title":"Journal of Cheminformatics"},{"key":"2026091723400788000_2026.09.08.750257v1.15","doi-asserted-by":"crossref","first-page":"1631","DOI":"10.1021\/acs.jcim.6b00359","article-title":"Prediction of activity cliffs using condensed graphs of reaction representations, descriptor recombination, support vector machine classification, and support vector regression","volume":"56","year":"2016","journal-title":"Journal of Chemical Information and Modeling"},{"key":"2026091723400788000_2026.09.08.750257v1.16","doi-asserted-by":"publisher","DOI":"10.1021\/ci7004093"},{"key":"2026091723400788000_2026.09.08.750257v1.17","doi-asserted-by":"publisher","DOI":"10.1021\/ci100050t"},{"key":"2026091723400788000_2026.09.08.750257v1.18","unstructured":"Rahaman, N. ; Baratin, A. ; Arpit, D. ; Draxler, F. ; Lin, M. ; Hamprecht, F. ; Bengio, Y. ; Courville, A. In International conference on machine learning, 2019, pp 5301\u20135310."},{"key":"2026091723400788000_2026.09.08.750257v1.19","doi-asserted-by":"crossref","first-page":"12978","DOI":"10.1021\/acs.jcim.5c01915","article-title":"MAPCliff-WMGR: Exploring Activity Cliffs in Molecular Activity Prediction Enhanced by Weighted Molecular Graph Representations","volume":"65","year":"2025","journal-title":"Journal of Chemical Information and Modeling"},{"key":"2026091723400788000_2026.09.08.750257v1.20","article-title":"A semi-supervised molecular learning framework for activity cliff estimation","year":"2026","journal-title":"arXiv preprint"},{"key":"2026091723400788000_2026.09.08.750257v1.21","doi-asserted-by":"crossref","unstructured":"Cui, C. ; Su, X. ; Zhang, Z. ; Velez-Arce, A. ; Wang, J. ; Shi, X. ; Zhang, Y. ; Wu, J. ; Chen, Y. Z. ; Zitnik, M. , et al. Activity-cliff awareness enables robust graph learning for molecular property prediction. Nature Communications 2026.","DOI":"10.26434\/chemrxiv-2023-5cz7s\/v4"},{"key":"2026091723400788000_2026.09.08.750257v1.22","first-page":"18661","article-title":"Supervised contrastive learning","volume":"33","year":"2020","journal-title":"Advances in neural information processing systems"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.64898\/2026.09.08.750257","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,18]],"date-time":"2026-09-18T06:40:19Z","timestamp":1789713619000},"score":1,"resource":{"primary":{"URL":"http:\/\/biorxiv.org\/lookup\/doi\/10.64898\/2026.09.08.750257"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,14]]},"references-count":22,"URL":"https:\/\/doi.org\/10.64898\/2026.09.08.750257","relation":{},"subject":[],"published":{"date-parts":[[2026,9,14]]},"subtype":"preprint"}}