{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,27]],"date-time":"2026-02-27T04:29:57Z","timestamp":1772166597764,"version":"3.50.1"},"reference-count":32,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2023,2,11]],"date-time":"2023-02-11T00:00:00Z","timestamp":1676073600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2023,2,11]],"date-time":"2023-02-11T00:00:00Z","timestamp":1676073600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["J Cheminform"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Artificial Intelligence is revolutionizing many aspects of the pharmaceutical industry. Deep learning models are now routinely applied to guide drug discovery projects leading to faster and improved findings, but there are still many tasks with enormous unrealized potential. One such task is the reaction yield prediction. Every year more than one fifth of all synthesis attempts result in product yields which are either zero or too low. This equates to chemical and human resources being spent on activities which ultimately do not progress the programs, leading to a triple loss when accounting for the cost of opportunity in time wasted. In this work we pre-train a BERT model on more than 16 million reactions from 4 different data sources, and fine tune it to achieve an uncertainty calibrated global yield prediction model. This model is an improvement upon state of the art not just from the increase in pre-train data but also by introducing a new embedding layer which solves a few limitations of SMILES and enables integration of additional information such as equivalents and molecule role into the reaction encoding, the model is called BERT Enriched Embedding (BEE). The model is benchmarked on an open-source dataset against a state-of-the-art synthesis focused BERT showing a near 20-point improvement in r2 score. The model is fine-tuned and tested on an internal company data benchmark, and a prospective study shows that the application of the model can reduce the total number of negative reactions (yield under 5%) ran in Janssen by at least 34%. Lastly, we corroborate the previous results through experimental validation, by directly deploying the model in an on-going drug discovery project and showing that it can also be used successfully as a reagent recommender due to its fast inference speed and reliable confidence estimation, a critical feature for industry application.<\/jats:p>","DOI":"10.1186\/s13321-023-00685-0","type":"journal-article","created":{"date-parts":[[2023,2,11]],"date-time":"2023-02-11T07:04:17Z","timestamp":1676099057000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":14,"title":["Global reactivity models are impactful in industrial synthesis applications"],"prefix":"10.1186","volume":"15","author":[{"given":"Paulo","family":"Neves","sequence":"first","affiliation":[]},{"given":"Kelly","family":"McClure","sequence":"additional","affiliation":[]},{"given":"Jonas","family":"Verhoeven","sequence":"additional","affiliation":[]},{"given":"Natalia","family":"Dyubankova","sequence":"additional","affiliation":[]},{"given":"Ramil","family":"Nugmanov","sequence":"additional","affiliation":[]},{"given":"Andrey","family":"Gedich","sequence":"additional","affiliation":[]},{"given":"Sairam","family":"Menon","sequence":"additional","affiliation":[]},{"given":"Zhicai","family":"Shi","sequence":"additional","affiliation":[]},{"given":"J\u00f6rg K.","family":"Wegner","sequence":"additional","affiliation":[]}],"member":"297","published-online":{"date-parts":[[2023,2,11]]},"reference":[{"issue":"16","key":"685_CR1","doi-asserted-by":"publisher","first-page":"8667","DOI":"10.1021\/acs.jmedchem.9b02120","volume":"63","author":"TJ Struble","year":"2020","unstructured":"Struble TJ et al (2020) Current and future roles of artificial intelligence in medicinal chemistry synthesis. J Med Chem 63(16):8667\u20138682. https:\/\/doi.org\/10.1021\/acs.jmedchem.9b02120","journal-title":"J Med Chem"},{"key":"685_CR2","doi-asserted-by":"publisher","first-page":"2","DOI":"10.1371\/journal.pcbi.1002380","volume":"8","author":"M Hartenfeller","year":"2012","unstructured":"Hartenfeller M et al (2012) \u201cDogs: reaction-driven de novo design of bioactive compounds. PLoS Comput Biol 8:2. https:\/\/doi.org\/10.1371\/journal.pcbi.1002380","journal-title":"PLoS Comput Biol"},{"issue":"1","key":"685_CR3","doi-asserted-by":"publisher","first-page":"384","DOI":"10.1038\/s41597-020-00727-4","volume":"7","author":"H Patel","year":"2020","unstructured":"Patel H et al (2020) SAVI, in silico generation of billions of easily synthesizable compounds through expert-system type rules. Sci Data 7(1):384. https:\/\/doi.org\/10.1038\/s41597-020-00727-4","journal-title":"Sci Data"},{"key":"685_CR4","doi-asserted-by":"crossref","unstructured":"M. Saebi et al. 2021. \u201cOn the Use of Real-World Datasets for Reaction Yield Prediction,\u201d pp. 1\u201324. https:\/\/doi.org\/10.26434\/chemrxiv-2021-2x06r-v3","DOI":"10.33774\/chemrxiv-2021-2x06r-v3"},{"key":"685_CR5","doi-asserted-by":"crossref","unstructured":"Schwaller P, Vaucher AC, Laino T, Reymond J-L (2021) \u201cPrediction of chemical reaction yields using deep learning,\u201d Mach Learn Sci Technol 2(1):015016. https:\/\/doi.org\/10.1088\/2632-2153\/abc81d","DOI":"10.1088\/2632-2153\/abc81d"},{"issue":"1","key":"685_CR6","first-page":"1","volume":"1","author":"M Saebi","year":"2021","unstructured":"Saebi M, Nan B, Herr J, Wahlers J, Wiest O (2021) Graph neural networks for predicting chemical reaction performance. ChemRxiv 1(1):1\u20134","journal-title":"ChemRxiv"},{"issue":"6","key":"685_CR7","doi-asserted-by":"publisher","first-page":"2198","DOI":"10.1039\/D0SC04823B","volume":"12","author":"Y Guan","year":"2021","unstructured":"Guan Y et al (2021) Regio-selectivity prediction with a machine-learned reaction representation and on-the-fly quantum mechanical descriptors. Chem Sci 12(6):2198\u20132208. https:\/\/doi.org\/10.1039\/D0SC04823B","journal-title":"Chem Sci"},{"issue":"1","key":"685_CR8","doi-asserted-by":"publisher","first-page":"3582","DOI":"10.1038\/s41598-017-02303-0","volume":"7","author":"G Skoraczy\u0144ski","year":"2017","unstructured":"Skoraczy\u0144ski G et al (2017) Predicting the outcomes of organic reactions via machine learning: are current descriptors sufficient? Sci Rep 7(1):3582. https:\/\/doi.org\/10.1038\/s41598-017-02303-0","journal-title":"Sci Rep"},{"issue":"1","key":"685_CR9","doi-asserted-by":"publisher","first-page":"1695","DOI":"10.1038\/s41467-021-21895-w","volume":"12","author":"DP Kov\u00e1cs","year":"2021","unstructured":"Kov\u00e1cs DP, McCorkindale W, Lee AA (2021) Quantitative interpretation explains machine learning models for chemical reaction prediction and uncovers bias. Nat Commun 12(1):1695. https:\/\/doi.org\/10.1038\/s41467-021-21895-w","journal-title":"Nat Commun"},{"issue":"11\u201312","key":"685_CR10","doi-asserted-by":"publisher","first-page":"615","DOI":"10.1002\/minf.201600073","volume":"35","author":"IV Tetko","year":"2016","unstructured":"Tetko IV, Engkvist O, Koch U, Reymond J-L, Chen H (2016) BIGCHEM: challenges and opportunities for big data analysis in chemistry. Mol Inform 35(11\u201312):615\u2013621. https:\/\/doi.org\/10.1002\/minf.201600073","journal-title":"Mol Inform"},{"issue":"6385","key":"685_CR11","doi-asserted-by":"publisher","first-page":"186","DOI":"10.1126\/science.aar5169","volume":"360","author":"DT Ahneman","year":"1979","unstructured":"Ahneman DT, Estrada JG, Lin S, Dreher SD, Doyle AG (1979) Predicting reaction performance in C-N cross-coupling using machine learning. Science 360(6385):186\u2013190. https:\/\/doi.org\/10.1126\/science.aar5169","journal-title":"Science"},{"issue":"8","key":"685_CR12","doi-asserted-by":"publisher","first-page":"1856","DOI":"10.1021\/acs.accounts.0c00770","volume":"54","author":"AM \u017bura\u0144ski","year":"2021","unstructured":"\u017bura\u0144ski AM, Martinez Alvarado JI, Shields BJ, Doyle AG (2021) Predicting reaction yields via supervised learning. Acc Chem Res 54(8):1856\u20131865. https:\/\/doi.org\/10.1021\/acs.accounts.0c00770","journal-title":"Acc Chem Res"},{"issue":"6","key":"685_CR13","doi-asserted-by":"publisher","first-page":"1379","DOI":"10.1016\/j.chempr.2020.02.017","volume":"6","author":"F Sandfort","year":"2020","unstructured":"Sandfort F, Strieth-Kalthoff F, K\u00fchnemund M, Beecks C, Glorius F (2020) A structure-based platform for predicting chemical reactivity. Chem 6(6):1379\u20131390. https:\/\/doi.org\/10.1016\/j.chempr.2020.02.017","journal-title":"Chem"},{"issue":"7844","key":"685_CR14","doi-asserted-by":"publisher","first-page":"89","DOI":"10.1038\/s41586-021-03213-y","volume":"590","author":"BJ Shields","year":"2021","unstructured":"Shields BJ et al (2021) Bayesian reaction optimization as a tool for chemical synthesis. Nature 590(7844):89\u201396. https:\/\/doi.org\/10.1038\/s41586-021-03213-y","journal-title":"Nature"},{"issue":"11","key":"685_CR15","doi-asserted-by":"publisher","first-page":"1465","DOI":"10.1021\/acscentsci.8b00357","volume":"4","author":"H Gao","year":"2018","unstructured":"Gao H, Struble TJ, Coley CW, Wang Y, Green WH, Jensen KF (2018) Using machine learning to predict suitable conditions for organic reactions. ACS Cent Sci 4(11):1465\u20131476. https:\/\/doi.org\/10.1021\/acscentsci.8b00357","journal-title":"ACS Cent Sci"},{"key":"685_CR16","unstructured":"Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez A, Kaiser L, Polosukhin I (2017) \"Attention is all you need\", In Advances in Neural Information Processing Systems. 5998\u20136008"},{"key":"685_CR17","unstructured":"Devlin J, Chang M-W, Lee K, Toutanova K (2018) \u201cBERT: Pre-training of Deep Bidirectional Transformers for Language Understanding\u201d arXiv:1810.04805. Retrieved from\nhttps:\/\/arxiv.org\/abs\/1810.04805"},{"issue":"1","key":"685_CR18","doi-asserted-by":"publisher","first-page":"015016","DOI":"10.1088\/2632-2153\/abc81d","volume":"2","author":"P Schwaller","year":"2021","unstructured":"Schwaller P, Vaucher AC, Laino T, Reymond J-L (2021) Prediction of chemical reaction yields using deep learning. Mach Learn Sci Technol 2(1):015016. https:\/\/doi.org\/10.1088\/2632-2153\/abc81d","journal-title":"Mach Learn Sci Technol"},{"key":"685_CR19","doi-asserted-by":"publisher","unstructured":"Schwaller P, Vaucher AC, Laino T, Reymond J-L. Data augmentation strategies to improve reaction yield predictions and estimate uncertainty. Theor Comp Chem. 2020.\nhttps:\/\/doi.org\/10.26434\/chemrxiv.13286741.v1","DOI":"10.26434\/chemrxiv.13286741.v1"},{"key":"685_CR20","doi-asserted-by":"publisher","DOI":"10.1039\/D1DD00006C","author":"D Probst","year":"2022","unstructured":"Probst D, Schwaller P, Reymond J-L (2022) Reaction classification and yield prediction using the differential reaction fingerprint DRFP. Digit Discov. https:\/\/doi.org\/10.1039\/D1DD00006C","journal-title":"Digit Discov"},{"key":"685_CR21","doi-asserted-by":"publisher","DOI":"10.1038\/s42256-020-00284-w","author":"P Schwaller","year":"2021","unstructured":"Schwaller P et al (2021) Mapping the space of chemical reactions using attention-based neural networks. Nat Mach Intell. https:\/\/doi.org\/10.1038\/s42256-020-00284-w","journal-title":"Nat Mach Intell"},{"key":"685_CR22","unstructured":"Lowe D (2017) Chemical reactions from US patents (1976-Sep2016).\nhttps:\/\/figshare.com\/articles\/Chemical_reactions_from_US_patents_1976-Sep2016_\/5104873"},{"issue":"12","key":"685_CR23","doi-asserted-by":"publisher","first-page":"2100119","DOI":"10.1002\/minf.202100119","volume":"40","author":"TR Gimadiev","year":"2021","unstructured":"Gimadiev TR et al (2021) Reaction data curation I: chemical structures and transformations standardization. Mol Inform 40(12):2100119. https:\/\/doi.org\/10.1002\/minf.202100119","journal-title":"Mol Inform"},{"key":"685_CR24","doi-asserted-by":"publisher","first-page":"15","DOI":"10.1126\/sciadv.abe4166","volume":"7","author":"P Schwaller","year":"2021","unstructured":"Schwaller P, Hoover B, Reymond J-L, Strobelt H, Laino T (2021) Extraction of organic chemistry grammar from unsupervised learning of chemical reactions. Sci Adv 7:15. https:\/\/doi.org\/10.1126\/sciadv.abe4166","journal-title":"Sci Adv"},{"issue":"6","key":"685_CR25","doi-asserted-by":"publisher","first-page":"2516","DOI":"10.1021\/acs.jcim.9b00102","volume":"59","author":"RI Nugmanov","year":"2019","unstructured":"Nugmanov RI et al (2019) CGRtools: python library for molecule, reaction, and condensed graph of reaction processing. J Chem Inf Model 59(6):2516\u20132521. https:\/\/doi.org\/10.1021\/acs.jcim.9b00102https:\/\/doi.org\/10.1021\/acs.jcim.9b00102","journal-title":"J Chem Inf Model"},{"key":"685_CR26","doi-asserted-by":"publisher","unstructured":"Guo C, Pleiss G, Sun Y, K. Q (2017) Weinberger, \u201cOn Calibration of Modern Neural Networks. https:\/\/doi.org\/10.48550\/arXiv.1706.04599","DOI":"10.48550\/arXiv.1706.04599"},{"issue":"8","key":"685_CR27","doi-asserted-by":"publisher","first-page":"1472","DOI":"10.3390\/rs13081472","volume":"13","author":"J Haas","year":"2021","unstructured":"Haas J, Rabus B (2021) Uncertainty estimation for deep learning-based segmentation of roads in synthetic aperture radar imagery. Remote Sens (Basel) 13(8):1472. https:\/\/doi.org\/10.3390\/rs13081472","journal-title":"Remote Sens (Basel)"},{"key":"685_CR28","unstructured":"M. Henne, A. Schwaiger, K. Roscher, and G. Weiss. (2020). \u201cBenchmarking Uncertainty Estimation Methods for Deep Learning With Safety-Related Metrics,\u201d 2020."},{"key":"685_CR29","doi-asserted-by":"publisher","DOI":"10.1021\/acscentsci.9b00576","author":"P Schwaller","year":"2019","unstructured":"Schwaller P et al (2019) Molecular transformer: a model for uncertainty-calibrated chemical reaction prediction. ACS Cent Sci. https:\/\/doi.org\/10.1021\/acscentsci.9b00576","journal-title":"ACS Cent Sci"},{"issue":"8","key":"685_CR30","doi-asserted-by":"publisher","first-page":"835","DOI":"10.1016\/j.envint.2010.02.011","volume":"36","author":"TJ van \u2019t Erve","year":"2010","unstructured":"van \u2019t Erve TJ, Rautiainen RH, Robertson LW, Luthe G (2010) Trimethylsilyldiazomethane: a safe non-explosive, cost effective and less-toxic reagent for phenol derivatization in GC applications. Environ Int 36(8):835\u2013842. https:\/\/doi.org\/10.1016\/j.envint.2010.02.011","journal-title":"Environ Int"},{"issue":"6","key":"685_CR31","doi-asserted-by":"publisher","first-page":"1213","DOI":"10.1021\/acs.oprd.9b00140","volume":"23","author":"SM Mennen","year":"2019","unstructured":"Mennen SM et al (2019) The evolution of high-throughput experimentation in pharmaceutical development and perspectives on the future. Org Process Res Dev 23(6):1213\u20131242. https:\/\/doi.org\/10.1021\/acs.oprd.9b00140","journal-title":"Org Process Res Dev"},{"issue":"7","key":"685_CR32","doi-asserted-by":"publisher","first-page":"1586","DOI":"10.1021\/acs.accounts.0c00760","volume":"54","author":"SD Dreher","year":"2021","unstructured":"Dreher SD, Krska SW (2021) Chemistry informer libraries: conception, early experience, and role in the future of cheminformatics. Acc Chem Res 54(7):1586\u20131596. https:\/\/doi.org\/10.1021\/acs.accounts.0c00760","journal-title":"Acc Chem Res"}],"updated-by":[{"DOI":"10.1186\/s13321-023-00705-z","type":"correction","label":"Correction","source":"publisher","updated":{"date-parts":[[2023,3,2]],"date-time":"2023-03-02T00:00:00Z","timestamp":1677715200000}}],"container-title":["Journal of Cheminformatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s13321-023-00685-0.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1186\/s13321-023-00685-0\/fulltext.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s13321-023-00685-0.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,3,3]],"date-time":"2023-03-03T02:04:38Z","timestamp":1677809078000},"score":1,"resource":{"primary":{"URL":"https:\/\/jcheminf.biomedcentral.com\/articles\/10.1186\/s13321-023-00685-0"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,2,11]]},"references-count":32,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2023,12]]}},"alternative-id":["685"],"URL":"https:\/\/doi.org\/10.1186\/s13321-023-00685-0","relation":{"has-preprint":[{"id-type":"doi","id":"10.26434\/chemrxiv-2022-5775s-v2","asserted-by":"object"},{"id-type":"doi","id":"10.26434\/chemrxiv-2022-5775s","asserted-by":"object"}]},"ISSN":["1758-2946"],"issn-type":[{"value":"1758-2946","type":"electronic"}],"subject":[],"published":{"date-parts":[[2023,2,11]]},"assertion":[{"value":"16 November 2022","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"22 January 2023","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"11 February 2023","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"2 March 2023","order":4,"name":"change_date","label":"Change Date","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"Correction","order":5,"name":"change_type","label":"Change Type","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"A Correction to this paper has been published:","order":6,"name":"change_details","label":"Change Details","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"https:\/\/doi.org\/10.1186\/s13321-023-00705-z","URL":"https:\/\/doi.org\/10.1186\/s13321-023-00705-z","order":7,"name":"change_details","label":"Change Details","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"Not applicable.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethics approval and consent to participate"}},{"value":"The authors\u00a0have\u00a0no conflicts of interest\u00a0to\u00a0declare.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}],"article-number":"20"}}