{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,2]],"date-time":"2026-08-02T23:45:29Z","timestamp":1785714329811,"version":"3.56.0"},"reference-count":60,"publisher":"IOP Publishing","issue":"2","license":[{"start":{"date-parts":[[2023,4,25]],"date-time":"2023-04-25T00:00:00Z","timestamp":1682380800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2023,4,25]],"date-time":"2023-04-25T00:00:00Z","timestamp":1682380800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"funder":[{"DOI":"10.13039\/100006233","name":"National Renewable Energy Laboratory","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100006233","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/100006134","name":"Office of Energy Efficiency and Renewable Energy","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100006134","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Mach. Learn.: Sci. Technol."],"published-print":{"date-parts":[[2023,6,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>A long-standing goal of machine-learning-based protein engineering is to accelerate the discovery of novel mutations that improve the function of a known protein. We introduce a sampling framework for evolving proteins <jats:italic>in silico<\/jats:italic> that supports mixing and matching a variety of unsupervised models, such as protein language models, and supervised models that predict protein function from sequence. By composing these models, we aim to improve our ability to evaluate unseen mutations and constrain search to regions of sequence space likely to contain functional proteins. Our framework achieves this without any model fine-tuning or re-training by constructing a product of experts distribution directly in discrete protein space. Instead of resorting to brute force search or random sampling, which is typical of classic directed evolution, we introduce a fast Markov chain Monte Carlo sampler that uses gradients to propose promising mutations. We conduct <jats:italic>in silico<\/jats:italic> directed evolution experiments on wide fitness landscapes and across a range of different pre-trained unsupervised models, including a 650 M parameter protein language model. Our results demonstrate an ability to efficiently discover variants with high evolutionary likelihood as well as estimated activity multiple mutations away from a wild type protein, suggesting our sampler provides a practical and effective new paradigm for machine-learning-based protein engineering.<\/jats:p>","DOI":"10.1088\/2632-2153\/accacd","type":"journal-article","created":{"date-parts":[[2023,4,6]],"date-time":"2023-04-06T08:59:21Z","timestamp":1680771561000},"page":"025014","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":18,"title":["Plug &amp; play directed evolution of proteins with gradient-based discrete MCMC"],"prefix":"10.1088","volume":"4","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-7846-5578","authenticated-orcid":true,"given":"Patrick","family":"Emami","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Aidan","family":"Perreault","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jeffrey","family":"Law","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"David","family":"Biagioni","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7928-3722","authenticated-orcid":false,"given":"Peter","family":"St. John","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"266","published-online":{"date-parts":[[2023,4,25]]},"reference":[{"key":"mlstaccacdbib1","article-title":"Model-based reinforcement learning for biological sequence design","author":"Angermueller","year":"2019"},{"key":"mlstaccacdbib2","first-page":"pp 324","article-title":"Population-based black-box optimization for biological sequence design","volume":"vol 119","author":"Angerm\u00fcller","year":"2020"},{"key":"mlstaccacdbib3","doi-asserted-by":"publisher","first-page":"125","DOI":"10.1021\/ar960017f","article-title":"Design by directed evolution","volume":"31","author":"Arnold","year":"1998","journal-title":"Acc. Chem. Res."},{"key":"mlstaccacdbib4","article-title":"Estimating or propagating gradients through stochastic neurons for conditional computation","author":"Bengio","year":"2013"},{"key":"mlstaccacdbib5","doi-asserted-by":"publisher","first-page":"389","DOI":"10.1038\/s41592-021-01100-y","article-title":"Low-n protein engineering with data-efficient deep learning","volume":"18","author":"Biswas","year":"2021","journal-title":"Nat. Methods"},{"key":"mlstaccacdbib6","article-title":"Design by adaptive sampling","author":"Brookes","year":"2018"},{"key":"mlstaccacdbib7","first-page":"pp 773","article-title":"Conditioning by adaptive sampling for robust design","volume":"vol 97","author":"Brookes","year":"2019"},{"key":"mlstaccacdbib8","article-title":"ReLSO: a transformer-based model for latent space optimization and generation of proteins","author":"Castro","year":"2022"},{"key":"mlstaccacdbib9","first-page":"14084","article-title":"Deep extrapolation for attribute-enhanced generation","volume":"34","author":"Chan","year":"2021","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"mlstaccacdbib10","article-title":"How to hallucinate functional proteins","author":"Costello","year":"2019"},{"key":"mlstaccacdbib11","doi-asserted-by":"crossref","DOI":"10.1101\/2021.11.09.467890","article-title":"Flip: benchmark tasks in fitness landscape inference for proteins","author":"Dallago","year":"2021"},{"key":"mlstaccacdbib12","article-title":"Plug and play language models: a simple approach to controlled text generation","author":"Dathathri","year":"2020"},{"key":"mlstaccacdbib13","article-title":"Autofocused oracles for model-based design","author":"Fannjiang","year":"2020"},{"key":"mlstaccacdbib14","doi-asserted-by":"crossref","DOI":"10.1038\/s41467-022-32007-7","article-title":"A deep unsupervised language model for protein design","author":"Ferruz","year":"2022"},{"key":"mlstaccacdbib15","doi-asserted-by":"crossref","DOI":"10.1101\/2021.12.22.473759","article-title":"Function-guided protein design by deep manifold sampling","author":"Gligorijevic","year":"2021"},{"key":"mlstaccacdbib16","doi-asserted-by":"publisher","first-page":"268","DOI":"10.1021\/acscentsci.7b00572","article-title":"Automatic chemical design using a data-driven continuous representation of molecules","volume":"4","author":"Gomez-Bombarelli","year":"2018","journal-title":"ACS Cent. Sci."},{"key":"mlstaccacdbib17","first-page":"pp 3831","article-title":"Oops I took a gradient: scalable sampling for discrete distributions","volume":"vol 139","author":"Grathwohl","year":"2021"},{"key":"mlstaccacdbib18","article-title":"Feedback GAN (FBGAN) for DNA: a novel feedback-loop architecture for optimizing protein functions","author":"Gupta","year":"2018"},{"key":"mlstaccacdbib19","first-page":"pp 312","article-title":"Adapting arbitrary normal mutation distributions in evolution strategies: the covariance matrix adaptation","author":"Hansen","year":"1996"},{"key":"mlstaccacdbib20","doi-asserted-by":"publisher","first-page":"97","DOI":"10.1093\/biomet\/57.1.97","article-title":"Monte Carlo sampling methods using Markov chains and their applications","volume":"57","author":"Hastings","year":"1970","journal-title":"Biometrika"},{"key":"mlstaccacdbib21","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pone.0090972","article-title":"Generating functional protein variants with variational autoencoders","volume":"17","author":"Hawkins-Hooker","year":"2021","journal-title":"PLoS Comput. Biol."},{"key":"mlstaccacdbib22","article-title":"RITA: a study on scaling up generative protein sequence models","author":"Hesslow","year":"2022"},{"key":"mlstaccacdbib23","doi-asserted-by":"publisher","first-page":"1771","DOI":"10.1162\/089976602760128018","article-title":"Training products of experts by minimizing contrastive divergence","volume":"14","author":"Hinton","year":"2002","journal-title":"Neural Comput."},{"key":"mlstaccacdbib24","doi-asserted-by":"crossref","DOI":"10.18653\/v1\/P18-1152","article-title":"Learning to write with cooperative discriminators","author":"Holtzman","year":"2018"},{"key":"mlstaccacdbib25","doi-asserted-by":"publisher","first-page":"128","DOI":"10.1038\/nbt.3769","article-title":"Mutation effects predicted from sequence co-variation","volume":"35","author":"Hopf","year":"2017","journal-title":"Nat. Biotechnol."},{"key":"mlstaccacdbib26","doi-asserted-by":"publisher","first-page":"1114","DOI":"10.1038\/s41587-021-01146-5","article-title":"Learning protein fitness models from evolutionary and assay-labeled data","volume":"40","author":"Hsu","year":"2022","journal-title":"Nat. Biotechnol."},{"key":"mlstaccacdbib27","first-page":"pp 9786","article-title":"Biological sequence design with gflownets","author":"Jain","year":"2022"},{"key":"mlstaccacdbib28","article-title":"Categorical reparameterization with gumbel-softmax","author":"Jang","year":"2016"},{"key":"mlstaccacdbib29","article-title":"Generating and designing DNA with deep generative models","author":"Killoran","year":"2017"},{"key":"mlstaccacdbib30","doi-asserted-by":"publisher","first-page":"523","DOI":"10.1016\/S0167-7799(97)01138-4","article-title":"Directed evolution of enzyme catalysts","volume":"15","author":"Kuchner","year":"1997","journal-title":"Trends Biotechnol."},{"key":"mlstaccacdbib31","article-title":"Model inversion networks for model-based optimization","author":"Kumar","year":"2020"},{"key":"mlstaccacdbib32","doi-asserted-by":"publisher","first-page":"729","DOI":"10.1016\/j.tibtech.2019.12.008","article-title":"Protein engineering for improving and diversifying natural product biosynthesis","volume":"38","author":"Li","year":"2020","journal-title":"Trends Biotechnol."},{"key":"mlstaccacdbib33","article-title":"Language models of protein sequences at the scale of evolution enable accurate structure prediction","author":"Lin","year":"2022"},{"key":"mlstaccacdbib34","doi-asserted-by":"publisher","first-page":"49","DOI":"10.1016\/j.cels.2020.05.007","article-title":"A generative neural network for maximizing fitness and diversity of synthetic DNA and protein sequences","volume":"11","author":"Linder","year":"2020","journal-title":"Cell Syst."},{"key":"mlstaccacdbib35","doi-asserted-by":"crossref","DOI":"10.1101\/2020.03.07.982272","article-title":"ProGen: language modeling for protein generation","author":"Madani","year":"2020"},{"key":"mlstaccacdbib36","article-title":"The concrete distribution: a continuous relaxation of discrete random variables","author":"Maddison","year":"2016"},{"key":"mlstaccacdbib37","doi-asserted-by":"crossref","DOI":"10.1101\/2021.07.09.450648","article-title":"Language models enable zero-shot prediction of the effects of mutations on protein function","author":"Meier","year":"2021"},{"key":"mlstaccacdbib38","doi-asserted-by":"publisher","first-page":"1087","DOI":"10.1063\/1.1699114","article-title":"Equation of state calculations by fast computing machines","volume":"21","author":"Metropolis","year":"1953","journal-title":"J. Chem. Phys."},{"key":"mlstaccacdbib39","first-page":"1","article-title":"Introductory Lectures on Convex Programming Volume I: Basic course","author":"Nesterov","year":"1998"},{"key":"mlstaccacdbib40","first-page":"pp 3510","article-title":"Plug & play generative networks: conditional iterative generation of images in latent space","author":"Nguyen","year":"2017"},{"key":"mlstaccacdbib41","article-title":"ProGen2: exploring the boundaries of protein language models","author":"Nijkamp","year":"2022"},{"key":"mlstaccacdbib42","first-page":"pp 16990","article-title":"Tranception: protein fitness prediction with autoregressive transformers and inference-time retrieval","author":"Notin","year":"2022"},{"key":"mlstaccacdbib43","article-title":"COLD decoding: energy-based constrained text generation with langevin dynamics","author":"Qin","year":"2022"},{"key":"mlstaccacdbib44","first-page":"pp 8844","article-title":"MSA transformer","author":"Rao","year":"2021"},{"key":"mlstaccacdbib45","doi-asserted-by":"publisher","first-page":"816","DOI":"10.1038\/s41592-018-0138-4","article-title":"Deep generative models of genetic variation capture the effects of mutations","volume":"15","author":"Riesselman","year":"2018","journal-title":"Nat. Methods"},{"key":"mlstaccacdbib46","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.2016239118","article-title":"Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences","volume":"118","author":"Rives","year":"2021","journal-title":"Proc. Natl Acad. Sci."},{"key":"mlstaccacdbib47","doi-asserted-by":"publisher","first-page":"2403","DOI":"10.1038\/s41467-021-22732-w","article-title":"Protein design and variant prediction using autoregressive generative models","volume":"12","author":"Shin","year":"2021","journal-title":"Nat. Commun."},{"key":"mlstaccacdbib48","article-title":"AdaLead: a simple and robust adaptive greedy search algorithm for sequence design","author":"Sinai","year":"2020"},{"key":"mlstaccacdbib49","article-title":"Discrete Langevin sampler via Wasserstein gradient flow","author":"Sun","year":"2022a"},{"key":"mlstaccacdbib50","article-title":"Path auxiliary proposal for mcmc in discrete space","author":"Sun","year":"2022b"},{"key":"mlstaccacdbib51","article-title":"Intriguing properties of neural networks","author":"Szegedy","year":"2014"},{"key":"mlstaccacdbib52","article-title":"Design-bench: benchmarks for data-driven offline model-based optimization","author":"Trabucco","year":"2022"},{"key":"mlstaccacdbib53","first-page":"pp 10358","article-title":"Conservative objective models for effective offline model-based optimization","volume":"vol 139","author":"Trabucco","year":"2021"},{"key":"mlstaccacdbib54","first-page":"pp 5484","article-title":"Non-identifiability and the blessings of misspecification in models of molecular fitness","volume":"vol 35","author":"Weinstein","year":"2022"},{"key":"mlstaccacdbib55","doi-asserted-by":"publisher","first-page":"18","DOI":"10.1016\/j.cbpa.2021.04.004","article-title":"Protein sequence design with deep generative models","volume":"65","author":"Wu","year":"2021","journal-title":"Curr. Opin. Chem. Biol."},{"key":"mlstaccacdbib56","doi-asserted-by":"crossref","DOI":"10.1101\/2022.05.19.492714","article-title":"Convolutions are competitive with transformers for protein sequence pretraining","author":"Yang","year":"2022"},{"key":"mlstaccacdbib57","doi-asserted-by":"publisher","first-page":"687","DOI":"10.1038\/s41592-019-0496-6","article-title":"Machine-learning-guided directed evolution for protein engineering","volume":"16","author":"Yang","year":"2019","journal-title":"Nat. Methods"},{"key":"mlstaccacdbib58","doi-asserted-by":"publisher","first-page":"852","DOI":"10.1080\/01621459.2019.1585255","article-title":"Informed proposals for local mcmc in discrete spaces","volume":"115","author":"Zanella","year":"2020","journal-title":"J. Am. Stat. Assoc."},{"key":"mlstaccacdbib59","article-title":"Unifying likelihood-free inference with black-box optimization and beyond","author":"Zhang","year":"2022a"},{"key":"mlstaccacdbib60","first-page":"pp 26375","article-title":"A Langevin-like sampler for discrete distributions","author":"Zhang","year":"2022b"}],"container-title":["Machine Learning: Science and Technology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/accacd","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/accacd\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/accacd","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/accacd\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/accacd\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/accacd\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/accacd\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/accacd\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,4,25]],"date-time":"2023-04-25T07:55:22Z","timestamp":1682409322000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/accacd"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,4,25]]},"references-count":60,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2023,4,25]]},"published-print":{"date-parts":[[2023,6,1]]}},"URL":"https:\/\/doi.org\/10.1088\/2632-2153\/accacd","relation":{},"ISSN":["2632-2153"],"issn-type":[{"value":"2632-2153","type":"electronic"}],"subject":[],"published":{"date-parts":[[2023,4,25]]},"assertion":[{"value":"Plug & play directed evolution of proteins with gradient-based discrete MCMC","name":"article_title","label":"Article Title"},{"value":"Machine Learning: Science and Technology","name":"journal_title","label":"Journal Title"},{"value":"paper","name":"article_type","label":"Article Type"},{"value":"\u00a9 2023 The Author(s). Published by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2022-12-16","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2023-03-29","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2023-04-25","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}