{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,24]],"date-time":"2026-06-24T15:58:59Z","timestamp":1782316739753,"version":"3.54.5"},"reference-count":51,"publisher":"Oxford University Press (OUP)","issue":"2","license":[{"start":{"date-parts":[[2020,8,14]],"date-time":"2020-08-14T00:00:00Z","timestamp":1597363200000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/"}],"funder":[{"name":"Spanish MINECO"},{"name":"FEDER Project","award":["TEC2016-80141-P"],"award-info":[{"award-number":["TEC2016-80141-P"]}]},{"DOI":"10.13039\/100013248","name":"FPI","doi-asserted-by":"publisher","award":["BES-2017-079792"],"award-info":[{"award-number":["BES-2017-079792"]}],"id":[{"id":"10.13039\/100013248","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2021,4,19]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Motivation<\/jats:title>\n                    <jats:p>Protein function prediction is a difficult bioinformatics problem. Many recent methods use deep neural networks to learn complex sequence representations and predict function from these. Deep supervised models require a lot of labeled training data which are not available for this task. However, a very large amount of protein sequences without functional labels is available.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>We applied an existing deep sequence model that had been pretrained in an unsupervised setting on the supervised task of protein molecular function prediction. We found that this complex feature representation is effective for this task, outperforming hand-crafted features such as one-hot encoding of amino acids, k-mer counts, secondary structure and backbone angles. Also, it partly negates the need for complex prediction models, as a two-layer perceptron was enough to achieve competitive performance in the third Critical Assessment of Functional Annotation benchmark. We also show that combining this sequence representation with protein 3D structure information does not lead to performance improvement, hinting that 3D structure is also potentially learned during the unsupervised pretraining.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Availability and implementation<\/jats:title>\n                    <jats:p>Implementations of all used models can be found at https:\/\/github.com\/stamakro\/GCN-for-Structure-and-Function.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Supplementary information<\/jats:title>\n                    <jats:p>Supplementary data are available at Bioinformatics online.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1093\/bioinformatics\/btaa701","type":"journal-article","created":{"date-parts":[[2020,8,12]],"date-time":"2020-08-12T07:42:36Z","timestamp":1597218156000},"page":"162-170","source":"Crossref","is-referenced-by-count":118,"title":["Unsupervised protein embeddings outperform hand-crafted sequence and structure features at predicting molecular function"],"prefix":"10.1093","volume":"37","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-3286-049X","authenticated-orcid":false,"given":"Amelia","family":"Villegas-Morcillo","sequence":"first","affiliation":[{"name":"Department of Signal Theory, Telematics and Communications, University of Granada , 18071 Granada, Spain"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3111-4268","authenticated-orcid":false,"given":"Stavros","family":"Makrodimitris","sequence":"additional","affiliation":[{"name":"Delft Bioinformatics Lab, Delft University of Technology , 2628XE Delft, The Netherlands"},{"name":"Keygene N.V. , 6708PW Wageningen, The Netherlands"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Roeland C H J","family":"van Ham","sequence":"additional","affiliation":[{"name":"Delft Bioinformatics Lab, Delft University of Technology , 2628XE Delft, The Netherlands"},{"name":"Keygene N.V. , 6708PW Wageningen, The Netherlands"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Angel M","family":"Gomez","sequence":"additional","affiliation":[{"name":"Department of Signal Theory, Telematics and Communications, University of Granada , 18071 Granada, Spain"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Victoria","family":"Sanchez","sequence":"additional","affiliation":[{"name":"Department of Signal Theory, Telematics and Communications, University of Granada , 18071 Granada, Spain"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Marcel J T","family":"Reinders","sequence":"additional","affiliation":[{"name":"Delft Bioinformatics Lab, Delft University of Technology , 2628XE Delft, The Netherlands"},{"name":"Leiden Computational Biology Center, Leiden University Medical Center , 2333ZC Leiden, The Netherlands"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2020,8,14]]},"reference":[{"key":"2023051511004188200_btaa701-B1","doi-asserted-by":"crossref","first-page":"1315","DOI":"10.1038\/s41592-019-0598-1","article-title":"Unified rational protein engineering with sequence-based deep representation learning","volume":"16","author":"Alley","year":"2019","journal-title":"Nat. Methods"},{"key":"2023051511004188200_btaa701-B2","doi-asserted-by":"crossref","first-page":"403","DOI":"10.1016\/S0022-2836(05)80360-2","article-title":"Basic local alignment search tool","volume":"215","author":"Altschul","year":"1990","journal-title":"J. Mol. Biol"},{"key":"2023051511004188200_btaa701-B3","doi-asserted-by":"crossref","first-page":"223","DOI":"10.1126\/science.181.4096.223","article-title":"Principles that govern the folding of protein chains","volume":"181","author":"Anfinsen","year":"1973","journal-title":"Science"},{"key":"2023051511004188200_btaa701-B4","volume-title":"Nat. Genet.","author":"Ashburner","year":"2000"},{"key":"2023051511004188200_btaa701-B5","first-page":"199","article-title":"The pros and cons of predicting protein contact maps","volume":"413","author":"Bartoli","year":"2007","journal-title":"Methods Mol. Biol"},{"key":"2023051511004188200_btaa701-B6","author":"Bepler","year":"2019"},{"key":"2023051511004188200_btaa701-B7","author":"Berman","year":"2000"},{"key":"2023051511004188200_btaa701-B8","doi-asserted-by":"crossref","first-page":"397","DOI":"10.1002\/prot.25832","article-title":"Machine learning techniques for protein function prediction","volume":"88","author":"Bonetta","year":"2020","journal-title":"Proteins Struct. Funct. Bioinf"},{"key":"2023051511004188200_btaa701-B9","doi-asserted-by":"crossref","first-page":"1732","DOI":"10.3390\/molecules22101732","article-title":"ProLanGO: protein function prediction using neural machine translation based on a recurrent neural network","volume":"22","author":"Cao","year":"2017","journal-title":"Molecules"},{"key":"2023051511004188200_btaa701-B10","doi-asserted-by":"crossref","first-page":"113","DOI":"10.1186\/1471-2105-8-113","article-title":"Improved residue contact prediction using support vector machines and a large feature set","volume":"8","author":"Cheng","year":"2007","journal-title":"BMC Bioinformatics"},{"key":"2023051511004188200_btaa701-B11","doi-asserted-by":"crossref","first-page":"i53","DOI":"10.1093\/bioinformatics\/btt228","article-title":"Information-theoretic evaluation of predicted ontological annotations","volume":"29","author":"Clark","year":"2013","journal-title":"Bioinformatics"},{"key":"2023051511004188200_btaa701-B12","doi-asserted-by":"crossref","first-page":"31865","DOI":"10.1038\/srep31865","article-title":"FFPred 3: feature-based function prediction for all Gene Ontology domains","volume":"6","author":"Cozzetto","year":"2016","journal-title":"Sci. Rep"},{"key":"2023051511004188200_btaa701-B13","author":"Devlin","year":"2018"},{"key":"2023051511004188200_btaa701-B14","first-page":"1422","author":"Doersch","year":"2015"},{"key":"2023051511004188200_btaa701-B37"},{"key":"2023051511004188200_btaa701-B15","doi-asserted-by":"crossref","DOI":"10.1186\/1471-2105-11-283","article-title":"Optimal contact definition for reconstruction of Contact Maps","volume":"11","author":"Duarte","year":"2010","journal-title":"BMC Bioinformatics"},{"key":"2023051511004188200_btaa701-B16","author":"Eddy","year":"2009"},{"key":"2023051511004188200_btaa701-B17","doi-asserted-by":"crossref","first-page":"e0198216","DOI":"10.1371\/journal.pone.0198216","article-title":"Predicting human protein function with multitask deep neural networks","volume":"13","author":"Fa","year":"2018","journal-title":"PLoS One"},{"key":"2023051511004188200_btaa701-B18","first-page":"6530","volume-title":"Advances in Neural Information Processing Systems","author":"Fout","year":"2017"},{"key":"2023051511004188200_btaa701-B19","doi-asserted-by":"crossref","first-page":"3150","DOI":"10.1093\/bioinformatics\/bts565","article-title":"CD-HIT: accelerated for clustering the next-generation sequencing data","volume":"28","author":"Fu","year":"2012","journal-title":"Bioinformatics"},{"key":"2023051511004188200_btaa701-B20","author":"Gidaris","year":"2018"},{"key":"2023051511004188200_btaa701-B21","author":"Gligorijevic","year":"2020"},{"key":"2023051511004188200_btaa701-B22","doi-asserted-by":"crossref","first-page":"723","DOI":"10.1186\/s12859-019-3220-8","article-title":"Modeling aspects of the language of life through transfer-learning protein sequences","volume":"20","author":"Heinzinger","year":"2019","journal-title":"BMC Bioinformatics"},{"key":"2023051511004188200_btaa701-B23","doi-asserted-by":"crossref","first-page":"184","DOI":"10.1186\/s13059-016-1037-6","article-title":"An expanded evaluation of protein function prediction methods shows an improvement in accuracy","volume":"17","author":"Jiang","year":"2016","journal-title":"Genome Biol"},{"key":"2023051511004188200_btaa701-B24","doi-asserted-by":"crossref","first-page":"999","DOI":"10.1093\/bioinformatics\/btu791","article-title":"MetaPSICOV: combining coevolution methods for accurate prediction of contacts and long range hydrogen bonding in proteins","volume":"31","author":"Jones","year":"2015","journal-title":"Bioinformatics"},{"key":"2023051511004188200_btaa701-B25","first-page":"2577","article-title":"Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features","volume":"22","author":"Kabsch","year":"1983","journal-title":"Biopolym. Orig. Res. Biomol"},{"key":"2023051511004188200_btaa701-B26","first-page":"730481","author":"Kane","year":"2019"},{"key":"2023051511004188200_btaa701-B27","doi-asserted-by":"crossref","first-page":"2848","DOI":"10.1073\/pnas.71.7.2848","article-title":"On some principles governing molecular evolution","volume":"71","author":"Kimura","year":"1974","journal-title":"Proc. Natl. Acad. Sci. USA"},{"key":"2023051511004188200_btaa701-B28","author":"Kingma","year":"2015"},{"key":"2023051511004188200_btaa701-B29","author":"Kipf","year":"2019"},{"key":"2023051511004188200_btaa701-B30","doi-asserted-by":"crossref","first-page":"422","DOI":"10.1093\/bioinformatics\/btz595","article-title":"DeepGOPlus: improved protein function prediction from sequence","volume":"36","author":"Kulmanov","year":"2020","journal-title":"Bioinformatics"},{"key":"2023051511004188200_btaa701-B31","doi-asserted-by":"crossref","first-page":"660","DOI":"10.1093\/bioinformatics\/btx624","article-title":"DeepGO: predicting protein functions from sequence and interactions using a deep ontology-aware classifier","volume":"34","author":"Kulmanov","year":"2018","journal-title":"Bioinformatics"},{"key":"2023051511004188200_btaa701-B32","author":"Liu","year":"2017"},{"key":"2023051511004188200_btaa701-B33","doi-asserted-by":"crossref","first-page":"i773","DOI":"10.1093\/bioinformatics\/bty585","article-title":"Learning structural motif representations for efficient protein structure search","volume":"34","author":"Liu","year":"2018","journal-title":"Bioinformatics"},{"key":"2023051511004188200_btaa701-B34","doi-asserted-by":"crossref","first-page":"2040","DOI":"10.1002\/jcc.23718","article-title":"Predicting backbone Ca angles and dihedrals from protein sequences by stacked sparse auto-encoder deep neural network","volume":"35","author":"Lyons","year":"2014","journal-title":"J. Comput. Chem"},{"key":"2023051511004188200_btaa701-B35","author":"Mathis","year":"2019"},{"key":"2023051511004188200_btaa701-B36","author":"McCann","year":"2017"},{"key":"2023051511004188200_btaa701-B38","author":"Pesquita","year":"2007"},{"key":"2023051511004188200_btaa701-B39","author":"Peters","year":"2018"},{"key":"2023051511004188200_btaa701-B40","doi-asserted-by":"crossref","first-page":"221","DOI":"10.1038\/nmeth.2340","article-title":"A large-scale evaluation of computational protein function prediction","volume":"10","author":"Radivojac","year":"2013","journal-title":"Nat. Methods"},{"key":"2023051511004188200_btaa701-B41","first-page":"9689","volume-title":"Advances in Neural Information Processing Systems","author":"Rao","year":"2019"},{"key":"2023051511004188200_btaa701-B42","author":"Rives","year":"2019"},{"key":"2023051511004188200_btaa701-B43","first-page":"1929","article-title":"Dropout: a simple way to prevent neural networks from overfitting","volume":"15","author":"Srivastava","year":"2014","journal-title":"J. Mach. Learn. Res"},{"key":"2023051511004188200_btaa701-B44","doi-asserted-by":"crossref","DOI":"10.1038\/s41598-019-43708-3","article-title":"DEEPred: automated protein function prediction with multi-task feed-forward deep neural networks","volume":"9,","author":"Sureyya Rifaioglu","year":"2019","journal-title":"Sci. Rep"},{"key":"2023051511004188200_btaa701-B45","doi-asserted-by":"crossref","first-page":"e1005324","DOI":"10.1371\/journal.pcbi.1005324","article-title":"Accurate De Novo prediction of protein contact map by ultra-deep learning model","volume":"13","author":"Wang","year":"2017","journal-title":"PLoS Comput. Biol"},{"key":"2023051511004188200_btaa701-B46","doi-asserted-by":"crossref","first-page":"e1000105","DOI":"10.1371\/journal.pcbi.1000105","article-title":"Local function conservation in sequence and structure space","volume":"4","author":"Weinhold","year":"2008","journal-title":"PLoS Comput. Biol"},{"key":"2023051511004188200_btaa701-B47","doi-asserted-by":"crossref","first-page":"233","DOI":"10.1006\/jmbi.2000.3550","article-title":"Assessing annotation transfer for genomics: quantifying the relations between protein sequence, structure and function through traditional and probabilistic scores","volume":"297","author":"Wilson","year":"2000","journal-title":"J. Mol. Biol"},{"key":"2023051511004188200_btaa701-B48","first-page":"610444","author":"Zamora-Resendiz","year":"2019"},{"key":"2023051511004188200_btaa701-B49","doi-asserted-by":"crossref","first-page":"e1007411","DOI":"10.1371\/journal.pcbi.1007411","article-title":"Detecting distant-homology protein structures by aligning deep neural-network based contact maps","volume":"15","author":"Zheng","year":"2019","journal-title":"PLoS Comput. Biol"},{"key":"2023051511004188200_btaa701-B50","doi-asserted-by":"crossref","first-page":"244","DOI":"10.1186\/s13059-019-1835-8","article-title":"The CAFA challenge reports improved protein function prediction and new functional annotations for hundreds of genes through experimental screens","volume":"20","author":"Zhou","year":"2019","journal-title":"Genome Biol"},{"key":"2023051511004188200_btaa701-B51","doi-asserted-by":"crossref","first-page":"3749","DOI":"10.1093\/bioinformatics\/btx514","article-title":"Improving protein fold recognition by extracting fold-specific features from predicted residue\u2013residue contacts","volume":"33","author":"Zhu","year":"2017","journal-title":"Bioinformatics"}],"container-title":["Bioinformatics"],"original-title":[],"language":"en","link":[{"URL":"http:\/\/academic.oup.com\/bioinformatics\/advance-article-pdf\/doi\/10.1093\/bioinformatics\/btaa701\/34127048\/btaa701.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/37\/2\/162\/50322152\/btaa701.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/37\/2\/162\/50322152\/btaa701.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,5,15]],"date-time":"2023-05-15T07:01:47Z","timestamp":1684134107000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article\/37\/2\/162\/5892762"}},"subtitle":[],"editor":[{"given":"Arne","family":"Elofsson","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2020,8,14]]},"references-count":51,"journal-issue":{"issue":"2","published-print":{"date-parts":[[2021,4,19]]}},"URL":"https:\/\/doi.org\/10.1093\/bioinformatics\/btaa701","relation":{"has-preprint":[{"id-type":"doi","id":"10.1101\/2020.04.07.028373","asserted-by":"object"}]},"ISSN":["1367-4803","1367-4811"],"issn-type":[{"value":"1367-4803","type":"print"},{"value":"1367-4811","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2021,1,15]]},"published":{"date-parts":[[2020,8,14]]}}}