{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,28]],"date-time":"2026-10-28T13:14:43Z","timestamp":1793193283104,"version":"4.5.0"},"reference-count":58,"publisher":"American Chemical Society (ACS)","issue":"3","license":[{"start":{"date-parts":[[2020,3,4]],"date-time":"2020-03-04T00:00:00Z","timestamp":1583280000000},"content-version":"vor","delay-in-days":58,"URL":"https:\/\/pubs.acs.org\/page\/policy\/authorchoice_termsofuse.html"}],"funder":[{"DOI":"10.13039\/501100003006","name":"Eidgen\u00f6ssische Technische Hochschule Z\u00fcrich","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100003006","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001711","name":"Schweizerischer Nationalfonds zur F\u00f6rderung der Wissenschaftlichen Forschung","doi-asserted-by":"publisher","award":["205321_182176"],"award-info":[{"award-number":["205321_182176"]}],"id":[{"id":"10.13039\/501100001711","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Novartis Forschungsstiftung"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2020,3,23]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Recurrent neural networks (RNNs) are able to generate de novo molecular designs using simplified molecular input line entry systems (SMILES) string representations of the chemical structure. RNN-based structure generation is usually performed unidirectionally, by growing SMILES strings from left to right. However, there is no natural start or end of a small molecule, and SMILES strings are intrinsically nonunivocal representations of molecular graphs. These properties motivate bidirectional structure generation. Here, bidirectional generative RNNs for SMILES-based molecule design are introduced. To this end, two established bidirectional methods were implemented, and a new method for SMILES string generation and data augmentation is introduced\u2500the bidirectional molecule design by alternate learning (BIMODAL). These three bidirectional strategies were compared to the unidirectional forward RNN approach for SMILES string generation, in terms of the (i) novelty, (ii) scaffold diversity, and (iii) chemical\u2013biological relevance of the computer-generated molecules. The results positively advocate bidirectional strategies for SMILES-based molecular de novo design, with BIMODAL showing superior results to the unidirectional forward RNN for most of the criteria in the tested conditions. The code of the methods and the pretrained models can be found at URL https:\/\/github.com\/ETHmodlab\/BIMODAL.<\/jats:p>","DOI":"10.1021\/acs.jcim.9b00943","type":"journal-article","created":{"date-parts":[[2020,1,6]],"date-time":"2020-01-06T12:06:40Z","timestamp":1578312400000},"page":"1175-1183","source":"Crossref","is-referenced-by-count":217,"title":["Bidirectional Molecule Generation with Recurrent Neural\nNetworks"],"prefix":"10.1021","volume":"60","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-8552-6615","authenticated-orcid":true,"given":"Francesca","family":"Grisoni","sequence":"first","affiliation":[{"name":"ETH Zurich , , Vladimir-Prelog-Weg 4 , ,","place":["Zurich, Switzerland, 8093"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Michael","family":"Moret","sequence":"additional","affiliation":[{"name":"ETH Zurich , , Vladimir-Prelog-Weg 4 , ,","place":["Zurich, Switzerland, 8093"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Robin","family":"Lingwood","sequence":"additional","affiliation":[{"name":"ETH Zurich , , Vladimir-Prelog-Weg 4 , ,","place":["Zurich, Switzerland, 8093"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6706-1084","authenticated-orcid":true,"given":"Gisbert","family":"Schneider","sequence":"additional","affiliation":[{"name":"ETH Zurich , , Vladimir-Prelog-Weg 4 , ,","place":["Zurich, Switzerland, 8093"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"316","published-online":{"date-parts":[[2020,1,6]]},"reference":[{"key":"2026101606212197700_cit1","doi-asserted-by":"publisher","first-page":"824","DOI":"10.1038\/nature03192","article-title":"Chemical\nSpace and Biology","volume":"432","author":"Dobson","year":"2004","journal-title":"Nature"},{"key":"2026101606212197700_cit2","doi-asserted-by":"publisher","first-page":"160","DOI":"10.1016\/s1359-6446(97)01163-x","article-title":"Virtual screening-an overview","volume":"3","author":"Walters","year":"1998","journal-title":"Drug Discov. Today"},{"key":"2026101606212197700_cit3","doi-asserted-by":"publisher","first-page":"997","DOI":"10.1021\/ci980083r","article-title":"Computer-Based\nStructure Determination: Then and Now","volume":"38","author":"Munk","year":"1998","journal-title":"J. Chem.\nInf. Comput. Sci."},{"key":"2026101606212197700_cit4","doi-asserted-by":"publisher","first-page":"342","DOI":"10.1021\/ci600423u","article-title":"Virtual Exploration\nof the Chemical Universe up to\n11 Atoms of C, N, O, F: Assembly of 26.4 Million Structures (110.9\nMillion Stereoisomers) and Analysis for New Ring Systems, Stereochemistry,\nPhysicochemical Properties, Compound Classes, and Drug Discovery","volume":"47","author":"Fink","year":"2007","journal-title":"J. Chem. Inf. Model."},{"key":"2026101606212197700_cit5","doi-asserted-by":"publisher","first-page":"413","DOI":"10.1021\/ci9502663","article-title":"Principles\nof the Generation of Constitutional\nand Configurational Isomers","volume":"36","author":"Wieland","year":"1996","journal-title":"J. Chem. Inf. Comput.\nSci."},{"key":"2026101606212197700_cit6","doi-asserted-by":"publisher","first-page":"111","DOI":"10.1002\/minf.200900038","article-title":"Exhaustive\nStructure Generation for\nInverse-QSPR\/QSAR","volume":"29","author":"Miyao","year":"2010","journal-title":"Mol. Inf."},{"key":"2026101606212197700_cit7","doi-asserted-by":"publisher","first-page":"286","DOI":"10.1021\/acs.jcim.5b00628","article-title":"Inverse QSPR\/QSAR Analysis for Chemical\nStructure Generation (from y to x)","volume":"56","author":"Miyao","year":"2016","journal-title":"J. Chem.\nInf. Model."},{"key":"2026101606212197700_cit8","doi-asserted-by":"publisher","first-page":"764","DOI":"10.1002\/minf.201400072","article-title":"Ring-System-Based\nExhaustive Structure\nGeneration for Inverse-QSPR\/QSAR","volume":"33","author":"Miyao","year":"2014","journal-title":"Mol. Inf."},{"key":"2026101606212197700_cit9","doi-asserted-by":"publisher","first-page":"656","DOI":"10.1021\/ci6005307","article-title":"Flux (2): Comparison of Molecular\nMutation and Crossover\nOperators for Ligand-Based de Novo Design","volume":"47","author":"Fechner","year":"2007","journal-title":"J.\nChem. Inf. Model."},{"key":"2026101606212197700_cit10","doi-asserted-by":"publisher","first-page":"543","DOI":"10.1016\/j.asoc.2014.09.042","article-title":"Evolutionary\nalgorithms for de novo\ndrug design\u2500A survey","volume":"27","author":"Devi","year":"2015","journal-title":"Appl. Soft Comput."},{"key":"2026101606212197700_cit11","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pcbi.1002380","article-title":"DOGS: Reaction-Driven\nde Novo Design of Bioactive Compounds","volume":"8","author":"Hartenfeller","year":"2012","journal-title":"PLoS\nComput. Biol."},{"key":"2026101606212197700_cit12","doi-asserted-by":"publisher","first-page":"307","DOI":"10.1038\/s42256-019-0067-7","article-title":"Automated\nde Novo Molecular Design by Hybrid Machine Intelligence and Rule-Driven\nChemical Synthesis","volume":"1","author":"Button","year":"2019","journal-title":"Nat. Mach. Intell."},{"key":"2026101606212197700_cit13","doi-asserted-by":"crossref","unstructured":"Rumelhart, D. E.; Hinton, G. E.; Williams, R. J.\n          Learning Internal\nRepresentations by Error Propagation, ICS-8506; California University San Diego La Jolla, Institute for Cognitive\nScience, 1985.","DOI":"10.21236\/ADA164453"},{"key":"2026101606212197700_cit14","doi-asserted-by":"publisher","first-page":"2554","DOI":"10.1073\/pnas.79.8.2554","article-title":"Neural\nNetworks and Physical Systems with Emergent Collective Computational\nAbilities","volume":"79","author":"Hopfield","year":"1982","journal-title":"Proc. Natl. Acad. Sci. U.S.A."},{"key":"2026101606212197700_cit15","article-title":"Adversarial Autoencoders","author":"Makhzani","year":"2015"},{"key":"2026101606212197700_cit16","doi-asserted-by":"publisher","first-page":"268","DOI":"10.1021\/acscentsci.7b00572","article-title":"Automatic\nChemical Design Using a Data-Driven Continuous Representation of Molecules","volume":"4","author":"Go\u0301mez-Bombarelli","year":"2018","journal-title":"ACS Cent. Sci."},{"key":"2026101606212197700_cit17","doi-asserted-by":"publisher","first-page":"1300","DOI":"10.1002\/cmdc.201800204","article-title":"Designing\nAnticancer Peptides by\nConstructive Machine Learning","volume":"13","author":"Grisoni","year":"2018","journal-title":"ChemMedChem"},{"key":"2026101606212197700_cit18","doi-asserted-by":"publisher","first-page":"1194","DOI":"10.1021\/acs.jcim.7b00690","article-title":"Reinforced\nAdversarial Neural Computer for de Novo Molecular Design","volume":"58","author":"Putin","year":"2018","journal-title":"J. Chem. Inf. Model."},{"key":"2026101606212197700_cit19","doi-asserted-by":"publisher","first-page":"4398","DOI":"10.1021\/acs.molpharmaceut.8b00839","article-title":"Entangled Conditional\nAdversarial Autoencoder for de Novo Drug Discovery","volume":"15","author":"Polykovskiy","year":"2018","journal-title":"Mol. Pharm."},{"key":"2026101606212197700_cit20","article-title":"Objective-Reinforced\nGenerative Adversarial Networks (ORGAN) for Sequence Generation Models","author":"Guimaraes","year":"2017"},{"key":"2026101606212197700_cit21","doi-asserted-by":"publisher","first-page":"120","DOI":"10.1021\/acscentsci.7b00512","article-title":"Generating Focused\nMolecule Libraries for Drug Discovery with Recurrent Neural Networks","volume":"4","author":"Segler","year":"2018","journal-title":"ACS Cent. Sci."},{"key":"2026101606212197700_cit22","doi-asserted-by":"publisher","first-page":"31","DOI":"10.1021\/ci00057a005","article-title":"SMILES, a\nChemical Language and Information System. 1. Introduction to Methodology\nand Encoding Rules","volume":"28","author":"Weininger","year":"1988","journal-title":"J. Chem. Inf. Comput. Sci."},{"key":"2026101606212197700_cit23","article-title":"In Silico Generation\nof Novel,\nDrug-like Chemical Matter Using the LSTM Neural Network","author":"Ertl","year":"2017"},{"key":"2026101606212197700_cit24","doi-asserted-by":"publisher","first-page":"1700111","DOI":"10.1002\/minf.201700111","article-title":"Generative Recurrent Networks for De Novo Drug Design","volume":"37","author":"Gupta","year":"2018","journal-title":"Mol. Inf."},{"key":"2026101606212197700_cit25","article-title":"Molecular Generation\nwith Recurrent Neural Networks (RNNs)","author":"Bjerrum","year":"2017"},{"key":"2026101606212197700_cit26","doi-asserted-by":"publisher","first-page":"875","DOI":"10.1021\/acs.jcim.6b00754","article-title":"Chemical\nSpace Mimicry for Drug Discovery","volume":"57","author":"Yuan","year":"2017","journal-title":"J. Chem. Inf.\nModel."},{"key":"2026101606212197700_cit27","doi-asserted-by":"publisher","first-page":"1096","DOI":"10.1021\/acs.jcim.8b00839","article-title":"GuacaMol: Benchmarking\nModels for de Novo Molecular Design","volume":"59","author":"Brown","year":"2019","journal-title":"J. Chem.\nInf. Model."},{"key":"2026101606212197700_cit28","doi-asserted-by":"publisher","first-page":"1735","DOI":"10.1162\/neco.1997.9.8.1735","article-title":"Long Short-Term\nMemory","volume":"9","author":"Hochreiter","year":"1997","journal-title":"Neural\nComput."},{"key":"2026101606212197700_cit29","doi-asserted-by":"publisher","first-page":"1700153","DOI":"10.1002\/minf.201700153","article-title":"De Novo Design\nof Bioactive Small Molecules by Artificial Intelligence","volume":"37","author":"Merk","year":"2018","journal-title":"Mol. Inf."},{"key":"2026101606212197700_cit30","doi-asserted-by":"publisher","first-page":"68","DOI":"10.1038\/s42004-018-0068-1","article-title":"Tuning Artificial\nIntelligence on the de Novo Design of Natural-Product-Inspired Retinoid\nX Receptor Modulators","volume":"1","author":"Merk","year":"2018","journal-title":"Commun. Chem."},{"key":"2026101606212197700_cit31","article-title":"Backward and Forward Language Modeling for Constrained\nSentence Generation","author":"Mou","year":"2015"},{"key":"2026101606212197700_cit32","unstructured":"Berglund, M.; Raiko, T.; Honkala, M.; Ka\u0308rkka\u0308inen, L.; Vetek, A.; Karhunen, J. T.\n          Bidirectional\nRecurrent Neural\nNetworks as Generative Models. In  Advances\nin Neural Information Processing Systems 28; Cortes, C., Lawrence, N. D., Lee, D. D., Sugiyama, M., Garnett, R., Eds.; Curran\nAssociates, Inc., 2015; pp 856\u2013864."},{"key":"2026101606212197700_cit33","volume-title":"Recurrent Neural Networks:\nDesign and Applications","author":"Jain","year":"1999","edition":"1"},{"key":"2026101606212197700_cit34","doi-asserted-by":"publisher","first-page":"107","DOI":"10.1142\/s0218488598000094","article-title":"The Vanishing\nGradient Problem During Learning Recurrent Neural Nets and Problem\nSolutions","volume":"06","author":"Hochreiter","year":"1998","journal-title":"Int. J. Uncertain. Fuzziness Knowledge-Based\nSyst."},{"key":"2026101606212197700_cit35","doi-asserted-by":"publisher","first-page":"339","DOI":"10.1016\/0893-6080(88)90007-x","article-title":"Generalization\nof Backpropagation with Application to a Recurrent Gas Market Model","volume":"1","author":"Werbos","year":"1988","journal-title":"Neural Netw."},{"key":"2026101606212197700_cit36","article-title":"Empirical Evaluation\nof Gated\nRecurrent Neural Networks on Sequence Modeling","author":"Chung","year":"2014"},{"key":"2026101606212197700_cit37","doi-asserted-by":"publisher","first-page":"2673","DOI":"10.1109\/78.650093","article-title":"Bidirectional Recurrent Neural Networks","volume":"45","author":"Schuster","year":"1997","journal-title":"IEEE Trans. Signal Process."},{"key":"2026101606212197700_cit38","doi-asserted-by":"crossref","unstructured":"Graves, A.; Jaitly, N.; Mohamed, A.\n          Hybrid Speech\nRecognition with Deep Bidirectional LSTM.  2013 IEEE Workshop on Automatic Speech Recognition and Understanding; IEEE: Olomouc,\nCzech Republic, 2013; pp 273\u2013278.","DOI":"10.1109\/ASRU.2013.6707742"},{"key":"2026101606212197700_cit39","doi-asserted-by":"publisher","first-page":"43","DOI":"10.1021\/acs.jcim.8b00263","article-title":"Conditional Molecular Design with Deep Generative Models","volume":"59","author":"Kang","year":"2019","journal-title":"J. Chem. Inf. Model."},{"key":"2026101606212197700_cit40","doi-asserted-by":"publisher","first-page":"1182","DOI":"10.1021\/acs.jcim.8b00751","article-title":"De Novo Molecular Design\nby Combining Deep Autoencoder\nRecurrent Neural Networks with Generative Topographic Mapping","volume":"59","author":"Sattarov","year":"2019","journal-title":"J. Chem. Inf. Model."},{"key":"2026101606212197700_cit41","doi-asserted-by":"crossref","unstructured":"Wang, C.; Yang, H.; Bartz, C.; Meinel, C.\n          Image Captioning with Deep Bidirectional\nLSTMs.  Proceedings of the 24th ACM International\nConference on Multimedia; MM \u201916; ACM: New York, NY, USA, 2016; pp 988\u2013997.","DOI":"10.1145\/2964284.2964299"},{"key":"2026101606212197700_cit42","doi-asserted-by":"publisher","first-page":"2009","DOI":"10.1093\/bioinformatics\/bty937","article-title":"Identifying\nAntimicrobial Peptides Using Word Embedding\nwith Deep Recurrent Neural Networks","volume":"35","author":"Hamid","year":"2019","journal-title":"Bioinformatics"},{"key":"2026101606212197700_cit43","article-title":"SMILES2Vec: An\nInterpretable General-Purpose Deep Neural Network for Predicting Chemical\nProperties","author":"Goh","year":"2017"},{"key":"2026101606212197700_cit44","unstructured":"ChEMBL22, 2016. 10.6019\/CHEMBL.Database.22."},{"key":"2026101606212197700_cit45","doi-asserted-by":"publisher","first-page":"2111","DOI":"10.1021\/acs.jcim.5b00543","article-title":"Get Your Atoms in\nOrder-An Open-Source\nImplementation of a Novel and Robust Molecular Canonicalization Algorithm","volume":"55","author":"Schneider","year":"2015","journal-title":"J. Chem. Inf. Model."},{"key":"2026101606212197700_cit46","unstructured":"RDKit:\nOpen-Source Cheminformatics, 2019, http:\/\/www.rdkit.org."},{"key":"2026101606212197700_cit47","article-title":"Adam: A Method for Stochastic\nOptimization","author":"Kingma","year":"2014"},{"key":"2026101606212197700_cit48","doi-asserted-by":"crossref","DOI":"10.26434\/chemrxiv.10119299.v1","article-title":"Generating Customized\nCompound\nLibraries for Drug Discovery with Machine Intelligence","author":"Moret","year":"2019"},{"key":"2026101606212197700_cit49","doi-asserted-by":"publisher","first-page":"128","DOI":"10.1038\/s42256-019-0030-7","article-title":"Mind and Machine\nin Drug Design","volume":"1","author":"Schneider","year":"2019","journal-title":"Nat. Mach. Intell."},{"key":"2026101606212197700_cit50","doi-asserted-by":"publisher","first-page":"1162","DOI":"10.1002\/qsar.200610091","article-title":"Scaffold-Hopping: How\nFar Can You\nJump?","volume":"25","author":"Schneider","year":"2006","journal-title":"QSAR Comb. Sci."},{"key":"2026101606212197700_cit51","doi-asserted-by":"publisher","first-page":"2887","DOI":"10.1021\/jm9602928","article-title":"The Properties of Known Drugs. 1. Molecular Frameworks","volume":"39","author":"Bemis","year":"1996","journal-title":"J. Med. Chem."},{"key":"2026101606212197700_cit52","doi-asserted-by":"publisher","first-page":"1736","DOI":"10.1021\/acs.jcim.8b00234","article-title":"Fre\u0301chet ChemNet Distance:\nA Metric for Generative\nModels for Molecules in Drug Discovery","volume":"58","author":"Preuer","year":"2018","journal-title":"J. Chem.\nInf. Model."},{"key":"2026101606212197700_cit53","doi-asserted-by":"crossref","unstructured":"Goh, G. B.; Siegel, C.; Vishnu, A.; Hodas, N.\n          Using Rule-Based Labels for\nWeak Supervised Learning: A ChemNet for Transferable Chemical Property\nPrediction.  Proceedings of the 24th ACM SIGKDD\nInternational Conference on Knowledge Discovery & Data Mining;\nKDD \u201918; ACM: New York, NY, USA, 2018; pp 302\u2013310.","DOI":"10.1145\/3219819.3219838"},{"key":"2026101606212197700_cit54","unstructured":"Simard, P.; Victorri, B.; LeCun, Y.; Denker, J.\n          Tangent Prop-a Formalism for\nSpecifying Selected Invariances in an Adaptive Network.  Advances in Neural Information Processing Systems; NIPS, 1992; pp 895\u2013903."},{"key":"2026101606212197700_cit55","article-title":"SMILES Enumeration\nas Data Augmentation for Neural Network Modeling of Molecules","author":"Bjerrum","year":"2017"},{"key":"2026101606212197700_cit56","doi-asserted-by":"publisher","first-page":"D1091","DOI":"10.1093\/nar\/gkt1068","article-title":"DrugBank\n4.0: Shedding New Light on Drug Metabolism","volume":"42","author":"Law","year":"2014","journal-title":"Nucleic Acids Res."},{"key":"2026101606212197700_cit57","article-title":"Molecular Sets\n(MOSES): A Benchmarking Platform for Molecular Generation Models","author":"Polykovskiy","year":"2018"},{"key":"2026101606212197700_cit58","doi-asserted-by":"publisher","first-page":"71","DOI":"10.1186\/s13321-019-0393-0","article-title":"Randomized SMILES strings\nimprove the quality of molecular generative models","volume":"11","author":"Aru\u0301s-Pous","year":"2019","journal-title":"J. Cheminf."}],"container-title":["Journal of Chemical Information and Modeling"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/pubs.acs.org\/doi\/pdf\/10.1021\/acs.jcim.9b00943","content-type":"application\/pdf","content-version":"vor","intended-application":"unspecified"},{"URL":"https:\/\/pubs.acs.org\/jcisd8\/article-pdf\/60\/3\/1175\/68673405\/acs.jcim.9b00943.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/pubs.acs.org\/jcisd8\/article-pdf\/60\/3\/1175\/68673405\/acs.jcim.9b00943.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,16]],"date-time":"2026-10-16T10:49:10Z","timestamp":1792147750000},"score":1,"resource":{"primary":{"URL":"https:\/\/pubs.acs.org\/jcisd8\/article\/60\/3\/1175\/862831\/Bidirectional-Molecule-Generation-with-Recurrent"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2020,1,6]]},"references-count":58,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2020,1,6]]},"published-print":{"date-parts":[[2020,3,23]]}},"URL":"https:\/\/doi.org\/10.1021\/acs.jcim.9b00943","relation":{},"ISSN":["1549-9596","1549-960X"],"issn-type":[{"value":"1549-9596","type":"print"},{"value":"1549-960X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2020,1,6]]}}}