{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,31]],"date-time":"2026-01-31T07:22:47Z","timestamp":1769844167591,"version":"3.49.0"},"update-to":[{"DOI":"10.1371\/journal.pcbi.1013004","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2025,5,6]],"date-time":"2025-05-06T00:00:00Z","timestamp":1746489600000}}],"reference-count":49,"publisher":"Public Library of Science (PLoS)","issue":"4","license":[{"start":{"date-parts":[[2025,4,22]],"date-time":"2025-04-22T00:00:00Z","timestamp":1745280000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"National Key Research and Development Project of China","award":["2021YFA1000103 to TS, 2021YFA1000102 to TS"],"award-info":[{"award-number":["2021YFA1000103 to TS, 2021YFA1000102 to TS"]}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62272479 to TS, 62372469 to TS"],"award-info":[{"award-number":["62272479 to TS, 62372469 to TS"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62202498 to SW"],"award-info":[{"award-number":["62202498 to SW"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Taishan Scholarship","award":["tstp20240506 to TS"],"award-info":[{"award-number":["tstp20240506 to TS"]}]},{"DOI":"10.13039\/501100007129","name":"Shandong Provincial Natural Science Foundation","doi-asserted-by":"crossref","award":["ZR2021QF023 to TS"],"award-info":[{"award-number":["ZR2021QF023 to TS"]}],"id":[{"id":"10.13039\/501100007129","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["www.ploscompbiol.org"],"crossmark-restriction":false},"short-container-title":["PLoS Comput Biol"],"abstract":"<jats:p>Accurate prediction of molecular properties is a key component of Artificial Intelligence-driven Drug Design (AIDD). Despite significant progress in improving these predictive models, balancing accuracy with computational complexity remains a challenge. Molecular topological and geometric features provide rich spatial information, crucial for improving prediction accuracy, but their extraction typically increases model complexity. To address this, we propose TGF-M (Topology-augmented Geometric Features for Molecular Property Prediction), a novel predictive model that optimizes feature extraction to enhance information capture and improve model accuracy, and reduces model complexity to lower computational cost. This approach enhances the model\u2019s ability to leverage both topological and geometric features without unnecessary complexity. On the re-segmented PCQM4Mv2 dataset, TGF-M performs remarkably, achieving a low mean absolute error (MAE) of 0.0647 in the HOMO-LUMO gap prediction task with only 6.4M parameters. Compared to two recent state-of-the-art models evaluated within a unified validation framework, TGF-M demonstrates comparable performance with less than one-tenth of the parameters. We conducted an in-depth analysis of TGF-M\u2019s chemical interpretability. The results further validate the method\u2019s effectiveness in leveraging complex molecular topology and geometry during model learning, underscoring its potential and advantages. The trained models and source code of TGF-M are publicly available at <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/github.com\/TiAW-Go\/TGF-M\" xlink:type=\"simple\">https:\/\/github.com\/TiAW-Go\/TGF-M<\/jats:ext-link>.<\/jats:p>","DOI":"10.1371\/journal.pcbi.1013004","type":"journal-article","created":{"date-parts":[[2025,4,22]],"date-time":"2025-04-22T20:02:22Z","timestamp":1745352142000},"page":"e1013004","update-policy":"https:\/\/doi.org\/10.1371\/journal.pcbi.corrections_policy","source":"Crossref","is-referenced-by-count":1,"title":["TGF-M: Topology-augmented geometric features enhance molecular property prediction"],"prefix":"10.1371","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0009-0006-0678-1516","authenticated-orcid":true,"given":"Wei","family":"He","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xu","family":"Tian","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xue","family":"Li","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Peifu","family":"Han","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Shuang","family":"Wang","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Lin","family":"Liu","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0130-3340","authenticated-orcid":true,"given":"Tao","family":"Song","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"340","published-online":{"date-parts":[[2025,4,22]]},"reference":[{"issue":"7","key":"pcbi.1013004.ref001","doi-asserted-by":"crossref","first-page":"722","DOI":"10.1021\/acsmedchemlett.5b00157","article-title":"Molecular property design: does everyone get it?","volume":"6","author":"PD Leeson","year":"2015","journal-title":"ACS Med Chem Lett"},{"key":"pcbi.1013004.ref002","unstructured":"Highest Occupied Molecular Orbital - an Overview ScienceDirect Topics. https:\/\/www.sciencedirect.com\/topics\/engineering\/highest-occupied-molecular-orbital"},{"issue":"1","key":"pcbi.1013004.ref003","doi-asserted-by":"crossref","first-page":"3","DOI":"10.1007\/s00894-016-3175-x","article-title":"Evaluating frontier orbital energy and HOMO\/LUMO gap with descriptors from density functional reactivity theory","volume":"23","author":"Y Huang","year":"2017","journal-title":"J Molecul Model"},{"issue":"2","key":"pcbi.1013004.ref004","doi-asserted-by":"crossref","first-page":"212","DOI":"10.1109\/TNB.2021.3056351","article-title":"Minimum free energy coding for DNA storage","volume":"20","author":"B Cao","year":"2021","journal-title":"IEEE Trans Nanobiosci"},{"key":"pcbi.1013004.ref005","first-page":"1","article-title":"Calculation of molecular structure and energy by force-field methods","volume-title":"Advances in physical organic chemistry","author":"NL Allinger","year":"1976"},{"issue":"15","key":"pcbi.1013004.ref006","doi-asserted-by":"crossref","first-page":"150901","DOI":"10.1063\/1.4704546","article-title":"Perspective on density functional theory","volume":"136","author":"K Burke","year":"2012","journal-title":"J Chem Phys"},{"issue":"12","key":"pcbi.1013004.ref007","doi-asserted-by":"crossref","first-page":"2326","DOI":"10.1021\/acs.jpclett.5b00831","article-title":"Machine learning predictions of molecular properties: accurate many-body potentials and nonlocality in chemical space","volume":"6","author":"K Hansen","year":"2015","journal-title":"J Phys Chem Lett"},{"issue":"4","key":"pcbi.1013004.ref008","doi-asserted-by":"crossref","first-page":"113699","DOI":"10.1016\/j.celrep.2024.113699","article-title":"Efficient data reconstruction: the bottleneck of large-scale application of DNA storage","volume":"43","author":"B Cao","year":"2024","journal-title":"Cell Reports"},{"key":"pcbi.1013004.ref009","article-title":"Advancements in molecular property prediction: a survey of single and multimodal approaches","author":"T Liyaqat","year":"2024"},{"key":"pcbi.1013004.ref010","article-title":"Molecular property prediction: recent trends in the era of artificial intelligence","author":"J Shen","year":"2019","journal-title":"Drug Discov Today Technol"},{"issue":"12","key":"pcbi.1013004.ref011","doi-asserted-by":"crossref","first-page":"103373","DOI":"10.1016\/j.drudis.2022.103373","article-title":"Deep learning methods for molecular representation and property prediction","volume":"27","author":"Z Li","year":"2022","journal-title":"Drug Discov Today"},{"key":"pcbi.1013004.ref012","first-page":"119","article-title":"Deep learning methods to help predict properties of molecules from SMILES. In: Kadoch M, Lu K, Ye F, Qian Y, editors. Proceedings of the International Symposium on Intelligent Computing and Networking 2024. vol. 1094. Cham: Springer","author":"G Bonilla-Caraballo","year":"2024"},{"key":"pcbi.1013004.ref013","doi-asserted-by":"crossref","unstructured":"Hirohara M, Saito Y, Koda Y, Sato K, Sakakibara Y. Convolutional neural network based on SMILES representation of compounds for detecting chemical motif. BMC Bioinformatics. 2018;19(S19):526. doi: 10.1186\/s12859-018-2523-5 30598075","DOI":"10.1186\/s12859-018-2523-5"},{"key":"pcbi.1013004.ref014","doi-asserted-by":"crossref","first-page":"65","DOI":"10.1016\/j.ymeth.2020.05.009","article-title":"The message passing neural networks for chemical property prediction on SMILES","volume":"179","author":"J Jo","year":"2020","journal-title":"Methods"},{"issue":"1","key":"pcbi.1013004.ref015","article-title":"MARPPI: boosting prediction of protein\u2013protein interactions with multi-scale architecture residual network","volume":"24","author":"X Li","year":"2023","journal-title":"Briefings Bioinform"},{"issue":"11","key":"pcbi.1013004.ref016","doi-asserted-by":"crossref","first-page":"1520","DOI":"10.1021\/acscentsci.8b00507","article-title":"PotentialNet for molecular property prediction","volume":"4","author":"EN Feinberg","year":"2018","journal-title":"ACS Cent Sci"},{"issue":"1","key":"pcbi.1013004.ref017","doi-asserted-by":"crossref","first-page":"474","DOI":"10.1186\/s12864-022-08687-2","article-title":"SDNN-PPI: self-attention with deep neural network effect on protein-protein interaction prediction","volume":"23","author":"X Li","year":"2022","journal-title":"BMC Genomics"},{"issue":"2","key":"pcbi.1013004.ref018","doi-asserted-by":"crossref","first-page":"127","DOI":"10.1038\/s42256-021-00438-4","article-title":"Geometry-enhanced molecular representation learning for property prediction","volume":"4","author":"X Fang","year":"2022","journal-title":"Nat Mach Intell"},{"issue":"6","key":"pcbi.1013004.ref019","doi-asserted-by":"crossref","DOI":"10.1093\/bioinformatics\/btad371","article-title":"3D graph contrastive learning for molecular property prediction","volume":"39","author":"K Moon","year":"2023","journal-title":"Bioinformatics"},{"issue":"3","key":"pcbi.1013004.ref020","doi-asserted-by":"crossref","first-page":"71","DOI":"10.1007\/s10044-024-01287-8","article-title":"3D-Mol: a novel contrastive learning framework for molecular property prediction with 3D information","volume":"27","author":"T Kuang","year":"2024","journal-title":"Pattern Anal Appl"},{"issue":"5","key":"pcbi.1013004.ref021","article-title":"PELMI: Realize robust DNA image storage under general errors via parity encoding and local mean iteration","volume":"25","author":"B Cao","year":"2024","journal-title":"Briefings Bioinform"},{"key":"pcbi.1013004.ref022","article-title":"Predicting mutation-disease associations through protein interactions via deep learning","author":"X Li","year":"2025","journal-title":"IEEE J Biomed Health Inform"},{"key":"pcbi.1013004.ref023","first-page":"2419","article-title":"Automated 3D pre-training for molecular property prediction. In: Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. Long Beach: ACM","author":"X Wang","year":"2023"},{"key":"pcbi.1013004.ref024","first-page":"20479","article-title":"3d Infomax improves GNNs for molecular property prediction. In: International Conference on Machine Learning. PMLR","author":"H St\u00e4rk","year":"2022"},{"key":"pcbi.1013004.ref025","unstructured":"Xu Z, Luo Y, Zhang X, Xu X, Xie Y, Liu M, et al. Molecule3D: a benchmark for predicting 3D geometries from molecular graphs; 2021."},{"key":"pcbi.1013004.ref026","article-title":"Semi-supervised classification with graph convolutional networks","author":"TN Kipf","year":"2016","journal-title":"arXiv preprint"},{"key":"pcbi.1013004.ref027","unstructured":"Xu K, Hu W, Leskovec J, Jegelka S. How powerful are graph neural networks? In: International Conference on Learning Representations; 2018."},{"key":"pcbi.1013004.ref028","first-page":"1263","article-title":"Neural message passing for quantum chemistry. In: International Conference on Machine Learning. PMLR","author":"J Gilmer","year":"2017"},{"key":"pcbi.1013004.ref029","unstructured":"Brossard R, Frigo O, Dehaene D. Graph convolutions that can finally model local structure; 2021."},{"key":"pcbi.1013004.ref030","first-page":"374","article-title":"Coatgin: marrying convolution and attention for graph-based molecule property prediction","author":"X Zhang","year":"2022"},{"key":"pcbi.1013004.ref031","unstructured":"Ying C, Cai T, Luo S, Zheng S, Ke G, He D, et al. Do transformers really perform badly for graph representation? Adv Neural Inf Process Syst. 2021;34:28877\u201388."},{"key":"pcbi.1013004.ref032","article-title":"Grpe: relative positional encoding for graph transformer","author":"W Park","year":"2022","journal-title":"arXiv preprint"},{"key":"pcbi.1013004.ref033","first-page":"1075","article-title":"PepPFN: protein-peptide binding residues prediction via pre-trained module-based Fourier Network","author":"X Li","year":"2024"},{"key":"pcbi.1013004.ref034","unstructured":"Hussain MS, Zaki MJ, Subramanian D. Global self-attention as a replacement for graph convolution; 2021. https:\/\/arxiv.org\/abs\/2108.03348v3"},{"key":"pcbi.1013004.ref035","unstructured":"Masters D, Dean J, Klaser K, Li Z, Maddrell-Mander S, Sanders A, et al. GPS++: an optimised hybrid MPNN\/transformer for molecular property prediction; 2022."},{"key":"pcbi.1013004.ref036","unstructured":"Luo S, Chen T, Xu Y, Zheng S, Liu TY, Wang L, et al. One transformer can understand both 2D & 3D molecular data. In: The Eleventh International Conference on Learning Representations; 2022."},{"key":"pcbi.1013004.ref037","article-title":"AEGNN-M: a 3D graph-spatial co-representation model for molecular property prediction","author":"L Cai","year":"2024","journal-title":"IEEE J Biomed Health Inform"},{"key":"pcbi.1013004.ref038","unstructured":"Lu S, Gao Z, He D, Zhang L, Ke G. Highly accurate quantum chemical property prediction with Uni-Mol+; 2023. https:\/\/arxiv.org\/abs\/2303.16982v2"},{"key":"pcbi.1013004.ref039","unstructured":"Hussain MS, Zaki MJ, Subramanian D. Triplet interaction improves graph transformers: accurate molecular graph learning with triplet graph transformers; 2024."},{"key":"pcbi.1013004.ref040","article-title":"Ogb-Lsc: a large-scale challenge for machine learning on graphs","author":"W Hu"},{"issue":"1","key":"pcbi.1013004.ref041","doi-asserted-by":"crossref","first-page":"111","DOI":"10.1063\/1.432807","article-title":"Evaluation of molecular integrals over Gaussian basis functions","volume":"65","author":"M Dupuis","year":"1976","journal-title":"J Chem Phys."},{"key":"pcbi.1013004.ref042","unstructured":"Molecular Topology - an Overview ScienceDirect Topics. https:\/\/www.sciencedirect.com\/topics\/chemistry\/molecular-topology"},{"key":"pcbi.1013004.ref043","doi-asserted-by":"crossref","unstructured":"Cao B, Wang K, Xie L, Zhang J, Zhao Y, Wang B, et al. PELMI: realize robust DNA image storage under general errors via parity encoding and local mean iteration. Briefings Bioinform. 2024;25(5):bbae463. doi: 10.1093\/bib\/bbae463 39288232","DOI":"10.1093\/bib\/bbae463"},{"key":"pcbi.1013004.ref044","doi-asserted-by":"crossref","unstructured":"Li C, Wang J, Niu Z, Yao J, Zeng X. A spatial-temporal gated attention module for molecular property prediction based on molecular geometry. Briefings Bioinform. 2021;22(5):bbab078. doi: 10.1093\/bib\/bbab078 33822856","DOI":"10.1093\/bib\/bbab078"},{"key":"pcbi.1013004.ref045","doi-asserted-by":"crossref","unstructured":"Cao B, Li X, Wang B, He T, Zheng Y, Zhang X, et al. Achieve handle level random access in encrypted DNA archival storage system via frequency dictionary mapping coding. bioRxiv. 2024; p. 2024\u201308.","DOI":"10.1101\/2024.08.15.608111"},{"key":"pcbi.1013004.ref046","unstructured":"Sch\u00f6lkopf B. The kernel trick for distances. Adv Neural Inf Process Syst. 2000;13:"},{"key":"pcbi.1013004.ref047","unstructured":"Veli\u010dkovi\u0107 P, Cucurull G, Casanova A, Romero A, Li\u00f2 P, Bengio Y. Graph attention networks; 2018."},{"key":"pcbi.1013004.ref048","unstructured":"Kim J, Nguyen TD, Min S, Cho S, Lee M, Lee H. Pure transformers are powerful graph learners; 2022."},{"issue":"10","key":"pcbi.1013004.ref049","doi-asserted-by":"crossref","first-page":"3448","DOI":"10.1021\/cr990357p","article-title":"Conjugated polymers and aromaticity","volume":"105","author":"M Kertesz","year":"2005","journal-title":"Chem Rev"}],"updated-by":[{"DOI":"10.1371\/journal.pcbi.1013004","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2025,5,6]],"date-time":"2025-05-06T00:00:00Z","timestamp":1746489600000}}],"container-title":["PLOS Computational Biology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1013004","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,5,6]],"date-time":"2025-05-06T17:58:35Z","timestamp":1746554315000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1013004"}},"subtitle":[],"editor":[{"given":"Shugang","family":"Zhang","sequence":"first","affiliation":[],"role":[{"role":"editor","vocabulary":"crossref"}]}],"short-title":[],"issued":{"date-parts":[[2025,4,22]]},"references-count":49,"journal-issue":{"issue":"4","published-online":{"date-parts":[[2025,4,22]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pcbi.1013004","relation":{"new_version":[{"id-type":"doi","id":"10.1371\/journal.pcbi.1013004","asserted-by":"object"}]},"ISSN":["1553-7358"],"issn-type":[{"value":"1553-7358","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,4,22]]}}}