{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,21]],"date-time":"2026-05-21T04:10:48Z","timestamp":1779336648932,"version":"3.51.4"},"reference-count":28,"publisher":"Oxford University Press (OUP)","issue":"5","license":[{"start":{"date-parts":[[2026,4,24]],"date-time":"2026-04-24T00:00:00Z","timestamp":1776988800000},"content-version":"vor","delay-in-days":4,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Wihuri Foundation"},{"name":"Jane and Aatos Erkko Foundation through the BIODESIGN"},{"DOI":"10.13039\/501100023750","name":"Helsinki Institute for Information Technology","doi-asserted-by":"crossref","id":[{"id":"10.13039\/501100023750","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/501100002341","name":"Research Council of Finland","doi-asserted-by":"publisher","award":["339421"],"award-info":[{"award-number":["339421"]}],"id":[{"id":"10.13039\/501100002341","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100002341","name":"Research Council of Finland","doi-asserted-by":"publisher","award":["345802"],"award-info":[{"award-number":["345802"]}],"id":[{"id":"10.13039\/501100002341","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Flagship Programme"},{"DOI":"10.13039\/100031478","name":"NextGenerationEU","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100031478","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,5,3]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Motivation<\/jats:title>\n                    <jats:p>Understanding chemical reactions requires bridging fine-grained molecular edits with broader semantic context. Reaction mechanisms are determined not only by local atom\u2013bond transformations but also by the global reaction class. However, most existing approaches treat these tasks separately or rely on external atom-mapping tools, introducing noise and limiting end-to-end learnability. We introduce MARCC (Mapping-Assisted Reaction Center and Classification), a multi-task graph neural network that jointly predicts atom mappings, reaction centers, and reaction classes within a unified architecture.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>MARCC integrates three key innovations: (i) a mapping-guided cross-attention mechanism that aligns reactants and products for local edit detection, (ii) a dual-graph design that explicitly reasons about bond-level transformations, and (iii) pooled product embeddings for global reaction classification. On the USPTO-50K benchmark, MARCC achieves state-of-the-art results when trained with both reactants and products, including 98.2% atom mapping accuracy, 99.1% Top-1 edit localization accuracy, and 97.2% reaction classification accuracy. Even under the products-only setting, MARCC delivers competitive performance comparable to specialized baselines. Ablation studies confirm the value of mapping-guided attention and multi-task supervision, which enhance both predictive accuracy and interpretability. By unifying atom-level alignment, local reactivity, and global classification, MARCC provides a structured and interpretable framework for reaction understanding. Beyond benchmarks, MARCC has the potential to support applications in reaction annotation, template discovery, and mechanism inference; with additional domain-specific modeling and data, it could be extended to biochemical domains such as enzyme-catalyzed transformations and metabolic pathway modeling.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Availability and implementation<\/jats:title>\n                    <jats:p>The source code and implementation details are available at https:\/\/github.com\/maryamastero\/MARCC and archived at https:\/\/doi.org\/10.5281\/zenodo.18500230.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1093\/bioinformatics\/btag193","type":"journal-article","created":{"date-parts":[[2026,4,18]],"date-time":"2026-04-18T11:12:24Z","timestamp":1776510744000},"source":"Crossref","is-referenced-by-count":0,"title":["A cross-attentive multi-task graph learning framework for chemical reaction modeling"],"prefix":"10.1093","volume":"42","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-4829-7627","authenticated-orcid":false,"given":"Maryam","family":"Astero","sequence":"first","affiliation":[{"name":"Department of Computer Science, Aalto University , Espoo 02150,","place":["Finland"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9828-6964","authenticated-orcid":false,"given":"Anchen","family":"Li","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Aalto University , Espoo 02150,","place":["Finland"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2024-7572","authenticated-orcid":false,"given":"Elena","family":"Casiraghi","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Aalto University , Espoo 02150,","place":["Finland"]},{"name":"Department of Computer Science, Universit\u00e0 degli Studi di Milano , Milan 20122,","place":["Italy"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0705-4314","authenticated-orcid":false,"given":"Juho","family":"Rousu","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Aalto University , Espoo 02150,","place":["Finland"]},{"name":"Department of Computer Science, University of Helsinki , Helsinki 00014,","place":["Finland"]}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"286","published-online":{"date-parts":[[2026,4,20]]},"reference":[{"key":"2026052023574620400_btag193-B1","first-page":"2623","author":"Akiba","year":"2019"},{"key":"2026052023574620400_btag193-B2","doi-asserted-by":"crossref","first-page":"46","DOI":"10.1186\/s13321-024-00841-0","article-title":"Learning symmetry-aware atom mapping in chemical reactions through deep graph matching","volume":"16","author":"Astero","year":"2024","journal-title":"J Cheminform"},{"key":"2026052023574620400_btag193-B3","doi-asserted-by":"crossref","first-page":"87","DOI":"10.1186\/s13321-025-01030-3","article-title":"Enhancing atom mapping with multitask learning and symmetry-aware deep graph matching","volume":"17","author":"Astero","year":"2025","journal-title":"J Cheminform"},{"key":"2026052023574620400_btag193-B4","article-title":"Algorithms for hyper-parameter optimization","volume":"24","author":"Bergstra","year":"2011","journal-title":"Adv Neural Inf Process Syst"},{"key":"2026052023574620400_btag193-B5","doi-asserted-by":"crossref","first-page":"2250","DOI":"10.1038\/s41467-024-46364-y","article-title":"Precise atom-to-atom mapping for organic reactions via human-in-the-loop machine learning","volume":"15","author":"Chen","year":"2024","journal-title":"Nat Commun"},{"key":"2026052023574620400_btag193-B6","doi-asserted-by":"crossref","first-page":"1237","DOI":"10.1021\/acscentsci.7b00355","article-title":"Computer-assisted retrosynthesis based on molecular similarity","volume":"3","author":"Coley","year":"2017","journal-title":"ACS Cent Sci"},{"key":"2026052023574620400_btag193-B7","article-title":"Retrosynthesis prediction with conditional graph logic network","volume":"32","author":"Dai","year":"2019","journal-title":"Adv Neural Inf Process Syst"},{"key":"2026052023574620400_btag193-B8","author":"Fabian","year":"2020"},{"key":"2026052023574620400_btag193-B9","author":"Fey","year":"2019"},{"key":"2026052023574620400_btag193-B10","doi-asserted-by":"crossref","first-page":"492","DOI":"10.1021\/ci00068a023","article-title":"Building and refining a knowledge base for synthetic organic chemistry via the methodology of inductive and deductive machine learning","volume":"30","author":"Gelernter","year":"1990","journal-title":"J Chem Inf Comput Sci"},{"key":"2026052023574620400_btag193-B11","doi-asserted-by":"crossref","first-page":"43","DOI":"10.1089\/cmb.2009.0216","article-title":"Computing atom mappings for biochemical reactions without subgraph isomorphism","volume":"18","author":"Heinonen","year":"2011","journal-title":"J Comput Biol"},{"key":"2026052023574620400_btag193-B12","first-page":"7482","author":"Kendall","year":"2018"},{"key":"2026052023574620400_btag193-B13","doi-asserted-by":"crossref","first-page":"110318","DOI":"10.1016\/j.patcog.2024.110318","article-title":"Rcsearcher: reaction center identification in retrosynthesis via deep q-learning","volume":"150","author":"Lan","year":"2024","journal-title":"Pattern Recognit"},{"key":"2026052023574620400_btag193-B14","author":"Landrum","year":"2006"},{"key":"2026052023574620400_btag193-B15","doi-asserted-by":"crossref","first-page":"btae558","DOI":"10.1093\/bioinformatics\/btae558","article-title":"Chemical reaction enhanced graph learning for molecule representation","volume":"40","author":"Li","year":"2024","journal-title":"Bioinformatics"},{"key":"2026052023574620400_btag193-B16","first-page":"2980","author":"Lin","year":"2017"},{"key":"2026052023574620400_btag193-B17","author":"Lowe","year":"2012"},{"key":"2026052023574620400_btag193-B18","first-page":"565","author":"Milletari","year":"2016"},{"key":"2026052023574620400_btag193-B19","doi-asserted-by":"crossref","first-page":"3307","DOI":"10.1021\/acs.jcim.2c00344","article-title":"Bidirectional graphormer for reactivity understanding: neural network trained to reaction atom-to-atom mapping task","volume":"62","author":"Nugmanov","year":"2022","journal-title":"J Chem Inf Model"},{"key":"2026052023574620400_btag193-B20","doi-asserted-by":"crossref","first-page":"91","DOI":"10.1039\/D1DD00006C","article-title":"Reaction classification and yield prediction using the differential reaction fingerprint drfp","volume":"1","author":"Probst","year":"2022","journal-title":"Digit Discov"},{"key":"2026052023574620400_btag193-B21","doi-asserted-by":"crossref","first-page":"521","DOI":"10.1023\/A:1021271615909","article-title":"Maximum common subgraph isomorphism algorithms for the matching of chemical structures","volume":"16","author":"Raymond","year":"2002","journal-title":"J Comput Aided Mol Des"},{"key":"2026052023574620400_btag193-B22","doi-asserted-by":"crossref","first-page":"39","DOI":"10.1021\/ci5006614","article-title":"Development of a novel fingerprint for chemical reactions and its application to large-scale reaction classification and similarity","volume":"55","author":"Schneider","year":"2015","journal-title":"J Chem Inf Model"},{"key":"2026052023574620400_btag193-B23","doi-asserted-by":"crossref","first-page":"2336","DOI":"10.1021\/acs.jcim.6b00564","article-title":"What\u2019s what: the (nearly) definitive guide to reaction role assignment","volume":"56","author":"Schneider","year":"2016","journal-title":"J Chem Inf Model"},{"key":"2026052023574620400_btag193-B24","doi-asserted-by":"crossref","first-page":"eabe4166","DOI":"10.1126\/sciadv.abe4166","article-title":"Extraction of organic chemistry grammar from unsupervised learning of chemical reactions","volume":"7","author":"Schwaller","year":"2021","journal-title":"Sci Adv"},{"key":"2026052023574620400_btag193-B25","doi-asserted-by":"crossref","first-page":"343","DOI":"10.2140\/pjm.1967.21.343","article-title":"Concerning nonnegative matrices and doubly stochastic matrices","volume":"21","author":"Sinkhorn","year":"1967","journal-title":"Pacific J Math"},{"key":"2026052023574620400_btag193-B26","first-page":"9405","article-title":"Learning graph models for retrosynthesis prediction","volume":"34","author":"Somnath","year":"2021","journal-title":"Adv Neural Inf Process Syst"},{"key":"2026052023574620400_btag193-B27","first-page":"12","article-title":"The reduction of a graph to canonical form and the algebra which appears therein","volume":"2","author":"Weisfeiler","year":"1968","journal-title":"nti, Series"},{"key":"2026052023574620400_btag193-B28","first-page":"11248","article-title":"Retroxpert: decompose retrosynthesis prediction like a chemist","volume":"33","author":"Yan","year":"2020","journal-title":"Adv Neural Inf Process Syst"}],"container-title":["Bioinformatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/bioinformatics\/advance-article-pdf\/doi\/10.1093\/bioinformatics\/btag193\/68144935\/btag193.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/42\/5\/btag193\/68144935\/btag193.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/42\/5\/btag193\/68144935\/btag193.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,21]],"date-time":"2026-05-21T03:58:27Z","timestamp":1779335907000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article\/doi\/10.1093\/bioinformatics\/btag193\/8661576"}},"subtitle":[],"editor":[{"given":"Jonathan","family":"Wren","sequence":"additional","affiliation":[],"role":[{"role":"editor","vocabulary":"crossref"}]}],"short-title":[],"issued":{"date-parts":[[2026,4,20]]},"references-count":28,"journal-issue":{"issue":"5","published-print":{"date-parts":[[2026,5,3]]}},"URL":"https:\/\/doi.org\/10.1093\/bioinformatics\/btag193","relation":{},"ISSN":["1367-4803","1367-4811"],"issn-type":[{"value":"1367-4803","type":"print"},{"value":"1367-4811","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,5]]},"published":{"date-parts":[[2026,4,20]]},"article-number":"btag193"}}