REVIEW RESEARCH 56.Smith,J.S..Isayev,O.Roitberg.A.E.Ani-1:an extensible neural network 80.Kim,E et al.Materials synthesis insights from scientific literature via text potential with DFT accuracy at force field computational cost Chem.Sci.8. extraction and machine learning.Chem.Mater.29,9436-9444(2017). 3192-3203(2017) 81.Jankowski,N.Duch,W.Grabczewski,K (eds)Meta-Learning in Computational 57.Bartok,A P,Payne,M.C.,Kondor,R.Csanyi,G.Gaussian approximation Intelligence(Springer.Berlin,2011). potentials:the accuracy of quantum mechanics,without the electrons.Phys. 82. Graves,A.Wayne,G.Danihelka,I.Neural Turing machines.Preprint at Rev.Lett.104.136403(2010). https:/axiw.og/abs/1410.5401(2014). In this study,machine learning is used to fit interatomic potentials that 83. Duan,Y.et al.One-shot imitation learning.Preprint at https://arxiv.org/ reproduce the total energy and energy derivatives from quantum-mechanical abs/1703.07326(201). calculations and enable accurate low-cost simulations. 84.Lake,B.M.,Salakhutdinov,R.Tenenbaum,J.B.Human-level concept learning 58.Handley,C.M.Popelier,P.L.A.Potential energy surfaces fitted by artificial through probabilistic program induction.Science 350,1332-1338(2015). neural networks.J.Phys.Chem.A 114.3371-3383(2010). 85.Ghiringhelli,L M.,Vybiral,J.,Levchenko,S.V..Draxl,C.Scheffler.M.Big data 59.Pulido,A.et al.Functional materials discovery using energy-structure-function of materials science:critical role of the descriptor.Phys.Rev.Lett.114,105503 maps.Nature543.657-664(2017). 2015. 60.Hill,J.et al.Materials science with large- -scale data and informatics:unlocking 86.Curtarolo,S.et al.The high-thro ughput highway to computational materials new opportunities.MRS Bull.41,399-409 (2016). design.Nat Mater.12,191-201(2013). 61. Kiselyova,N.N.,Gladun,V.P.Vashchenko,N.D.Computational materials 87.Seko,A,Togo,A Tanaka,I.in Nanoinformatics(ed.Tanaka,L.)3-23(Springer. design using artificial intelligence methods.J.Alloys Compd.279,8-13(1998). Singapore,2018). 62.Saal.J.E..Kirklin,S.Aykol.M.,Meredig.B.Wolverton,C.Materials design and 88.Duvenaud,D.et al.Convolutional raphs for lean molecular discovery with high-throughput density functional theory:the open quantum fingerprints.Preprint at https://arxiv.org/abs/1509.09292(2015) materials database (OQMD).JOM 65,1501-1509(2013). 89 Steane,A.Quantum computing.Rep.Prog.Phys.61,117 (1998). 63.Pilania,G.,Wang,C.,Jiang,X,Rajasekaran,S.Ramprasad,R.Accelerating 0. Harrow,A.W.Hassidim,A.Lloyd,S.Quantum algorithm for linear systems of materials property predictions using machine learning.Sci.Rep.3,2810 equations.Phys Rev.Lett.103,150502(2009). (2013). Aspuru-Guzik,A,Dutoi,A.D.,Love,P.J.Head-Gordon,M.Simulated quantum 64.Hautier,G.,Fischer,C.C..Jain,A.,Mueller,T.Ceder,G.Finding nature's missing computation of molecular energies.Science 309.1704-1707(2005). ternary oxide compounds using machine leaming and density functional In an early application of quantum computing to molecular problems,a theory.Chem.Mater.22,3762-3767(2010). quantum algorithm that scales linearly with the number of basis functions is In an early example of harnessing materials databases,information on known demonstrated for calculating properties of chemical interest. compounds is used to construct a machine-learning model to predict the 2 Reiher,M.Wiebe.N.Svore,KM.Wecker,D.&Troyer,M.Elucidating reaction viability of previously unreported chemistries. mechanisms on quantum computers.Proc.Natl Acad.Sci.USA 114, 65.Walsh,A.The quest for new functionality.Nat Chem.7,274-275(2015) 7555-7560(2017). 66.Davies,D.W.et al.Computational screening of all stoichiometric inorganic 93.Dunjko,V.,Taylor,J.M.Briegel,H.J.Quantum-enhanced machine learning. materials.Chem 1,617-627(2016). 67.Franceschetti,A Zunger,A.The inverse band-structure problem of finding an Phys.Rev.Lett.117,130501(2016. 94 Biamonte,J.et al.Quantum machine learning.Nature 549,195-202(2017) atomic configuration with given electronic properties.Nature 402,60-63 95.Schmidt,M.Lipson,H.Distilling free-form natural laws from experimental (1999) data.Science 324,81-85(2009) 68.Kuhn,C.Beratan,D.N.Inverse strategies for molecular design.J.Phys.Chem. 96.Rudy,S.H..Brunton,S.L.,Proctor,J.L.Kutz,J.N.Data-driven discovery of 100.10595-10599(1996) partial differential equations.Sci.Adv.3,e1602614(2017). 69.Oliynyk,A.O.et al.High-throughput machine-learning-driven synthesis of 97 Domingos,P.The Master Algorithm (Basic Books,New York,2015). full-Heusler compounds.Chem.Mater.28,7324-7331 (2016). 98.Coudert,F.-X.Reproducible research in computational chemistry of materials. 70.Legrain,F.,Carrete.J..van Roekeghem,A,Madsen,G.K.H.Mingo,N. Chem.Mater.29,2615-2617(2017). Materials screening for the discovery of new half-heuslers:machine learing 99.Tetko,I.V.,Maran,U.Tropsha,A.Public (Q)SAR services,integrated modeling versus ab initio methods.J.Phys.Chem.B 122,625-632(2018) environments,and model repositories on the web:state of the art and 71.Moot,T.et al.Material informatics driven design and experimental validation of perspectives for future development.Mol.Inform.36,1600082(2017) lead titanate as an aqueous solar photocathode.Mater.Discov.6,9-16(2016). 72.Faber,F.A,Lindmaa,A.,Von Lilienfeld,O.A Armiento,R.Machine learning Acknowledgements This work was supported by the EPSRC(grant numbers energies of 2 million elpasolite (ABC2D6)crystals.Phys.Rev.Lett.117,135502 EP/M009580/1,EP/K016288/1 and EP/L016354/1),the Royal Society and (2016). the Leverhulme Trust.O.l.acknowledges support from DOD-ONR(N00014-16. 73.Oprea,T.I.Tropsha.A.Target,chemical and bioactivity databases- 1-2311)and an Eshelman Institute for Innovation award. integration is key.Drug Discov.Today.Technol.3,357-365(2006). 74.Sterling.T.Irwin,J.J.ZINC 15-ligand discovery for everyone.J.Chem.Inf. Reviewer information Nature thanks F.-X.Coudert,M.Waller and the other Mode1.55,2324-2337(2015). anonymous reviewer(s)for their contribution to the peer review of this work 75.Tropsha,A.Best practices for QSAR model development,validation,and exploitation.Mol.Inform.29,476-488(2010). Author contributions All authors contributed equally to the design.writing and 76.Hansch,C.Fujita,T.p-o-analysis.A method for the correlation of biological editing of the manuscript. activity and chemical structure.J.Am.Chem.Soc.86,1616-1626(1964). 77.Goodfellow,L J.et al.Generative adversarial networks.Preprint at https://arxiv. Competing interests The authors declare no competing interests org/abs/1406.2661(2014). 78.Guimaraes,G.L Sanchez-Lengeling,B.Outeiral,C.Farias,P.LC.&Aspuru- Additional information Guzik,A.Objective-reinforced generative adversarial networks (ORGAN)for Reprints and permissions information is available at http://www.nature.com/ sequence generation models.Preprint at https://arxiv.org/abs/1705.10843 reprints. (2017). Correspondence and requests for materials should be addressed to O.I.or AW 79.Fleuren.W.W.M.Alkema.W.Application of text mining in the biomedical Publisher's note:Springer Nature remains neutral with regard to jurisdictional domain.Methods 74,97-106(2015). claims in published maps and institutional affiliations. 26 JULY 2018I VOL 559 I NATURE I 555 2018 Springer Nature Limited.All rights reserved.Review RESEARCH 56. Smith, J. S., Isayev, O. & Roitberg, A. E. Ani-1: an extensible neural network potential with DFT accuracy at force feld computational cost. Chem. Sci. 8, 3192–3203 (2017). 57. Bartók, A. P., Payne, M. C., Kondor, R. & Csányi, G. Gaussian approximation potentials: the accuracy of quantum mechanics, without the electrons. Phys. Rev. Lett. 104, 136403 (2010). In this study, machine learning is used to ft interatomic potentials that reproduce the total energy and energy derivatives from quantum-mechanical calculations and enable accurate low-cost simulations. 58. Handley, C. M. & Popelier, P. L. A. Potential energy surfaces ftted by artifcial neural networks. J. Phys. Chem. A 114, 3371–3383 (2010). 59. Pulido, A. et al. Functional materials discovery using energy–structure–function maps. Nature 543, 657–664 (2017). 60. Hill, J. et al. Materials science with large-scale data and informatics: unlocking new opportunities. MRS Bull. 41, 399–409 (2016). 61. Kiselyova, N. N., Gladun, V. P. & Vashchenko, N. D. Computational materials design using artifcial intelligence methods. J. Alloys Compd. 279, 8–13 (1998). 62. Saal, J. E., Kirklin, S., Aykol, M., Meredig, B. & Wolverton, C. Materials design and discovery with high-throughput density functional theory: the open quantum materials database (OQMD). JOM 65, 1501–1509 (2013). 63. Pilania, G., Wang, C., Jiang, X., Rajasekaran, S. & Ramprasad, R. Accelerating materials property predictions using machine learning. Sci. Rep. 3, 2810 (2013). 64. Hautier, G., Fischer, C. C., Jain, A., Mueller, T. & Ceder, G. Finding nature's missing ternary oxide compounds using machine learning and density functional theory. Chem. Mater. 22, 3762–3767 (2010). In an early example of harnessing materials databases, information on known compounds is used to construct a machine-learning model to predict the viability of previously unreported chemistries. 65. Walsh, A. The quest for new functionality. Nat. Chem. 7, 274–275 (2015). 66. Davies, D. W. et al. Computational screening of all stoichiometric inorganic materials. Chem 1, 617–627 (2016). 67. Franceschetti, A. & Zunger, A. The inverse band-structure problem of fnding an atomic confguration with given electronic properties. Nature 402, 60–63 (1999). 68. Kuhn, C. & Beratan, D. N. Inverse strategies for molecular design. J. Phys. Chem. 100, 10595–10599 (1996). 69. Oliynyk, A. O. et al. High-throughput machine-learning-driven synthesis of full-Heusler compounds. Chem. Mater. 28, 7324–7331 (2016). 70. Legrain, F., Carrete, J., van Roekeghem, A., Madsen, G. K. H. & Mingo, N. Materials screening for the discovery of new half-heuslers: machine learning versus ab initio methods. J. Phys. Chem. B 122, 625–632 (2018). 71. Moot, T. et al. Material informatics driven design and experimental validation of lead titanate as an aqueous solar photocathode. Mater. Discov. 6, 9–16 (2016). 72. Faber, F. A., Lindmaa, A., Von Lilienfeld, O. A. & Armiento, R. Machine learning energies of 2 million elpasolite (ABC2D6) crystals. Phys. Rev. Lett. 117, 135502 (2016). 73. Oprea, T. I. & Tropsha, A. Target, chemical and bioactivity databases – integration is key. Drug Discov. Today. Technol. 3, 357–365 (2006). 74. Sterling, T. & Irwin, J. J. ZINC 15 – ligand discovery for everyone. J. Chem. Inf. Model. 55, 2324–2337 (2015). 75. Tropsha, A. Best practices for QSAR model development, validation, and exploitation. Mol. Inform. 29, 476–488 (2010). 76. Hansch, C. & Fujita, T. p-σ-π analysis. A method for the correlation of biological activity and chemical structure. J. Am. Chem. Soc. 86, 1616–1626 (1964). 77. Goodfellow, I. J. et al. Generative adversarial networks. Preprint at https://arxiv. org/abs/1406.2661 (2014). 78. Guimaraes, G. L., Sanchez-Lengeling, B., Outeiral, C., Farias, P. L. C. & Aspuru￾Guzik, A. Objective-reinforced generative adversarial networks (ORGAN) for sequence generation models. Preprint at https://arxiv.org/abs/1705.10843 (2017). 79. Fleuren, W. W. M. & Alkema, W. Application of text mining in the biomedical domain. Methods 74, 97–106 (2015). 80. Kim, E. et al. Materials synthesis insights from scientifc literature via text extraction and machine learning. Chem. Mater. 29, 9436–9444 (2017). 81. Jankowski, N., Duch, W. & Grabczewski, K. (eds) ̧ Meta-Learning in Computational Intelligence (Springer, Berlin, 2011). 82. Graves, A., Wayne, G. & Danihelka, I. Neural Turing machines. Preprint at https://arxiv.org/abs/1410.5401 (2014). 83. Duan, Y. et al. One-shot imitation learning. Preprint at https://arxiv.org/ abs/1703.07326 (2017). 84. Lake, B. M., Salakhutdinov, R. & Tenenbaum, J. B. Human-level concept learning through probabilistic program induction. Science 350, 1332–1338 (2015). 85. Ghiringhelli, L. M., Vybiral, J., Levchenko, S. V., Draxl, C. & Schefer, M. Big data of materials science: critical role of the descriptor. Phys. Rev. Lett. 114, 105503 (2015). 86. Curtarolo, S. et al. The high-throughput highway to computational materials design. Nat. Mater. 12, 191–201 (2013). 87. Seko, A., Togo, A. & Tanaka, I. in Nanoinformatics (ed. Tanaka, I.) 3–23 (Springer, Singapore, 2018). 88. Duvenaud, D. et al. Convolutional networks on graphs for learning molecular fngerprints. Preprint at https://arxiv.org/abs/1509.09292 (2015). 89. Steane, A. Quantum computing. Rep. Prog. Phys. 61, 117 (1998). 90. Harrow, A. W., Hassidim, A. & Lloyd, S. Quantum algorithm for linear systems of equations. Phys. Rev. Lett. 103, 150502 (2009). 91. Aspuru-Guzik, A., Dutoi, A. D., Love, P. J. & Head-Gordon, M. Simulated quantum computation of molecular energies. Science 309, 1704–1707 (2005). In an early application of quantum computing to molecular problems, a quantum algorithm that scales linearly with the number of basis functions is demonstrated for calculating properties of chemical interest. 92. Reiher, M., Wiebe, N., Svore, K. M., Wecker, D. & Troyer, M. Elucidating reaction mechanisms on quantum computers. Proc. Natl Acad. Sci. USA 114, 7555–7560 (2017). 93. Dunjko, V., Taylor, J. M. & Briegel, H. J. Quantum-enhanced machine learning. Phys. Rev. Lett. 117, 130501 (2016). 94. Biamonte, J. et al. Quantum machine learning. Nature 549, 195–202 (2017). 95. Schmidt, M. & Lipson, H. Distilling free-form natural laws from experimental data. Science 324, 81–85 (2009). 96. Rudy, S. H., Brunton, S. L., Proctor, J. L. & Kutz, J. N. Data-driven discovery of partial diferential equations. Sci. Adv. 3, e1602614 (2017). 97. Domingos, P. The Master Algorithm (Basic Books, New York, 2015). 98. Coudert, F.-X. Reproducible research in computational chemistry of materials. Chem. Mater. 29, 2615–2617 (2017). 99. Tetko, I. V., Maran, U. & Tropsha, A. Public (Q)SAR services, integrated modeling environments, and model repositories on the web: state of the art and perspectives for future development. Mol. Inform. 36, 1600082 (2017). Acknowledgements This work was supported by the EPSRC (grant numbers EP/M009580/1, EP/K016288/1 and EP/L016354/1), the Royal Society and the Leverhulme Trust. O.I. acknowledges support from DOD-ONR (N00014-16- 1-2311) and an Eshelman Institute for Innovation award. Reviewer information Nature thanks F.-X. Coudert, M. Waller and the other anonymous reviewer(s) for their contribution to the peer review of this work. Author contributions All authors contributed equally to the design, writing and editing of the manuscript. Competing interests The authors declare no competing interests. Additional information Reprints and permissions information is available at http://www.nature.com/ reprints. Correspondence and requests for materials should be addressed to O.I. or A.W. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 26 J U LY 2018 | V OL 559 | NATUR E | 555 © 2018 Springer Nature Limited. All rights reserved