{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,4]],"date-time":"2026-09-04T14:28:14Z","timestamp":1788532094448,"version":"build-2803163510"},"reference-count":49,"publisher":"eLife Sciences Publications, Ltd","license":[{"start":{"date-parts":[[2026,9,4]],"date-time":"2026-09-04T00:00:00Z","timestamp":1788480000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"},{"start":{"date-parts":[[2026,9,4]],"date-time":"2026-09-04T00:00:00Z","timestamp":1788480000000},"content-version":"am","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"},{"start":{"date-parts":[[2026,9,4]],"date-time":"2026-09-04T00:00:00Z","timestamp":1788480000000},"content-version":"tdm","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Schmidt Science Fellows"},{"DOI":"10.13039\/100015539","name":"Australian Government","doi-asserted-by":"crossref","award":["Research Training Program"],"award-info":[{"award-number":["Research Training Program"]}],"id":[{"id":"10.13039\/100015539","id-type":"DOI","asserted-by":"crossref"}]},{"award":["Research Training Program"],"award-info":[{"award-number":["Research Training Program"]}],"id":[{"id":"https:\/\/ror.org\/0314h5y94","id-type":"ROR","asserted-by":"publisher"}]}],"content-domain":{"domain":["elifesciences.org"],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>Enzymes present a sustainable alternative to traditional chemical industries, drug synthesis, and bioremediation applications. Because catalytic residues are the key amino acids that drive enzyme function, their accurate prediction facilitates enzyme function prediction. Sequence similarity-based approaches such as BLAST are fast but require previously annotated homologues. Machine-learning (ML) approaches aim to overcome this limitation; however, current gold-standard ML-based methods require high-quality 3D structures limiting their application to large datasets. To address these challenges, we developed Squidly, a sequence-only tool that leverages contrastive representation learning with a biology-informed, rationally designed pairing scheme to distinguish catalytic from non-catalytic residues using per-token Protein Language Model embeddings. Squidly surpasses state-of-the-art ML annotation methods in catalytic residue prediction while remaining sufficiently fast to enable wide-scale screening of databases. We ensemble Squidly with BLAST to provide an efficient tool that annotates catalytic residues with high precision and recall for both in- and out-of-distribution sequences.<\/jats:p>","DOI":"10.7554\/elife.108186","type":"journal-article","created":{"date-parts":[[2025,10,2]],"date-time":"2025-10-02T15:31:33Z","timestamp":1759419093000},"update-policy":"https:\/\/doi.org\/10.7554\/elife.108186","source":"Crossref","is-referenced-by-count":2,"title":["Squidly harnesses enzyme functional hierarchy and contrastive learning to efficiently predict catalytic residues from sequence"],"prefix":"10.7554","volume":"14","author":[{"ORCID":"https:\/\/orcid.org\/0009-0005-9574-3468","authenticated-orcid":true,"given":"William JF","family":"Rieger","sequence":"first","affiliation":[{"id":[{"id":"https:\/\/ror.org\/00rqy9422","id-type":"ROR","asserted-by":"publisher"}],"name":"School of Chemistry and Molecular Biology, University of Queensland","place":["Queensland, Australia"]}],"role":[{"role":"author","vocabulary":"crossref"},{"role":"conceptualization","vocabulary":"credit"},{"role":"data-curation","vocabulary":"credit"},{"role":"formal-analysis","vocabulary":"credit"},{"role":"investigation","vocabulary":"credit"},{"role":"methodology","vocabulary":"credit"},{"role":"software","vocabulary":"credit"},{"role":"visualization","vocabulary":"credit"},{"role":"writing-original-draft","vocabulary":"credit"}]},{"given":"Mikael","family":"Bod\u00e9n","sequence":"additional","affiliation":[{"id":[{"id":"https:\/\/ror.org\/00rqy9422","id-type":"ROR","asserted-by":"publisher"}],"name":"School of Chemistry and Molecular Biology, University of Queensland","place":["Queensland, Australia"]}],"role":[{"role":"author","vocabulary":"crossref"},{"role":"conceptualization","vocabulary":"credit"},{"role":"funding-acquisition","vocabulary":"credit"},{"role":"project-administration","vocabulary":"credit"},{"role":"supervision","vocabulary":"credit"},{"role":"writing-review-editing","vocabulary":"credit"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4027-364X","authenticated-orcid":true,"given":"Frances","family":"Arnold","sequence":"additional","affiliation":[{"id":[{"id":"https:\/\/ror.org\/05dxps055","id-type":"ROR","asserted-by":"publisher"}],"name":"Division of Chemistry and Chemical Engineering, California Institute of Technology","place":["Pasadena, United States"]}],"role":[{"role":"author","vocabulary":"crossref"},{"role":"conceptualization","vocabulary":"credit"},{"role":"funding-acquisition","vocabulary":"credit"},{"role":"project-administration","vocabulary":"credit"},{"role":"supervision","vocabulary":"credit"},{"role":"writing-review-editing","vocabulary":"credit"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1331-8192","authenticated-orcid":true,"given":"Ariane","family":"Mora","sequence":"additional","affiliation":[{"id":[{"id":"https:\/\/ror.org\/05dxps055","id-type":"ROR","asserted-by":"publisher"}],"name":"Division of Chemistry and Chemical Engineering, California Institute of Technology","place":["Pasadena, United States"]}],"role":[{"role":"author","vocabulary":"crossref"},{"role":"corresponding-author","vocabulary":"crossref"},{"role":"conceptualization","vocabulary":"credit"},{"role":"data-curation","vocabulary":"credit"},{"role":"formal-analysis","vocabulary":"credit"},{"role":"investigation","vocabulary":"credit"},{"role":"methodology","vocabulary":"credit"},{"role":"resources","vocabulary":"credit"},{"role":"software","vocabulary":"credit"},{"role":"validation","vocabulary":"credit"},{"role":"writing-original-draft","vocabulary":"credit"}]}],"member":"4374","published-online":{"date-parts":[[2026,9,4]]},"reference":[{"key":"bib1","doi-asserted-by":"publisher","first-page":"96","DOI":"10.1038\/s41592-025-02975-x","type":"journal-article","article-title":"Atom-level enzyme active site scaffolding using RFdiffusion2","volume":"23","author":"Ahern","year":"2026","journal-title":"Nature Methods"},{"key":"bib2","doi-asserted-by":"publisher","first-page":"403","DOI":"10.1016\/S0022-2836(05)80360-2","type":"journal-article","article-title":"Basic local alignment search tool","volume":"215","author":"Altschul","year":"1990","journal-title":"Journal of Molecular Biology"},{"key":"bib3","doi-asserted-by":"publisher","first-page":"105","DOI":"10.1016\/s0022-2836(02)01036-7","type":"journal-article","article-title":"Analysis of catalytic residues in enzyme active sites","volume":"324","author":"Bartlett","year":"2002","journal-title":"Journal of Molecular Biology"},{"key":"bib4","doi-asserted-by":"publisher","DOI":"10.1101\/2024.10.10.615955","type":"preprint","article-title":"Chai-1: decoding the molecular interactions of life","volume-title":"bioRxiv","author":"Boitreaud","year":"2024"},{"key":"bib5","doi-asserted-by":"publisher","first-page":"59","DOI":"10.1038\/nmeth.3176","type":"journal-article","article-title":"Fast and sensitive protein alignment using DIAMOND","volume":"12","author":"Buchfink","year":"2015","journal-title":"Nature Methods"},{"key":"bib6","doi-asserted-by":"publisher","DOI":"10.1186\/1471-2105-8-153","type":"journal-article","article-title":"How accurate and statistically robust are catalytic site predictions based on closeness centrality?","volume":"8","author":"Chea","year":"2007","journal-title":"BMC Bioinformatics"},{"key":"bib7","doi-asserted-by":"publisher","first-page":"539","DOI":"10.1109\/CVPR.2005.202","type":"conference-paper","article-title":"Learning a Similarity Metric Discriminatively, with Application to Face Verification","author":"Chopra","year":"2005","unstructured":"Chopra S, Hadsell R, LeCun Y. 2005. Learning a Similarity Metric Discriminatively, with Application to Face Verification. 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR\u201905)."},{"key":"bib8","doi-asserted-by":"publisher","first-page":"D523","DOI":"10.1093\/nar\/gkac1052","type":"journal-article","article-title":"UniProt: the universal protein knowledgebase in 2023","volume":"51","author":"Consortium","year":"2022","journal-title":"Nucleic Acids Research"},{"key":"bib9","doi-asserted-by":"publisher","first-page":"1289","DOI":"10.1016\/S0022-2836(02)01451-1","type":"journal-article","article-title":"Automatic methods for predicting functionally important residues","volume":"326","author":"del Sol Mesa","year":"2003","journal-title":"Journal of Molecular Biology"},{"key":"bib10","doi-asserted-by":"publisher","first-page":"163","DOI":"10.1093\/bioinformatics\/19.1.163","type":"journal-article","article-title":"ConSurf: identification of functional regions in proteins by surface-mapping of phylogenetic information","volume":"19","author":"Glaser","year":"2003","journal-title":"Bioinformatics"},{"key":"bib11","doi-asserted-by":"publisher","first-page":"993","DOI":"10.1109\/34.58871","type":"journal-article","article-title":"Neural network ensembles","volume":"12","author":"Hansen","year":"1990","journal-title":"IEEE Transactions on Pattern Analysis and Machine Intelligence"},{"key":"bib12","doi-asserted-by":"publisher","DOI":"10.1101\/2023.07.23.550085","type":"preprint","article-title":"Bilingual Language Model for Protein Sequence and Structure","volume-title":"bioRxiv","author":"Heinzinger","year":"2024"},{"key":"bib13","doi-asserted-by":"publisher","first-page":"583","DOI":"10.1038\/s41586-021-03819-2","type":"journal-article","article-title":"Highly accurate protein structure prediction with AlphaFold","volume":"596","author":"Jumper","year":"2021","journal-title":"Nature"},{"key":"bib14","first-page":"21798","type":"book-chapter","article-title":"Hard negative mixing for contrastive learning","volume-title":"Advances in Neural Information Processing Systems","author":"Kalantidis","year":"2020"},{"key":"bib15","first-page":"18661","type":"book-chapter","article-title":"Supervised contrastive learning","volume-title":"Advances in Neural Information Processing Systems","author":"Khosla","year":"2020"},{"key":"bib16","doi-asserted-by":"publisher","DOI":"10.1101\/2023.11.15.567263","type":"preprint","article-title":"Semantic search using protein large language models detects class II microcins in bacterial genomes","volume-title":"bioRxiv","author":"Kulikova","year":"2023"},{"key":"bib17","type":"conference-paper","article-title":"Simple and scalable predictive uncertainty estimation using deep ensembles","author":"Lakshminarayanan","year":"2017","unstructured":"Lakshminarayanan B, Pritzel A, Blundell C. 2017. Simple and scalable predictive uncertainty estimation using deep ensembles. Advances in Neural Information Processing Systems."},{"key":"bib18","doi-asserted-by":"publisher","first-page":"193907","DOI":"10.1109\/ACCESS.2020.3031549","type":"journal-article","article-title":"Contrastive representation learning: a framework and review","volume":"8","author":"Le-Khac","year":"2020","journal-title":"IEEE Access"},{"key":"bib19","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2212.00552","type":"preprint","article-title":"An effective deployment of contrastive learning in multi-label text classification","volume-title":"arXiv","author":"Lin","year":"2022"},{"key":"bib20","doi-asserted-by":"publisher","first-page":"1123","DOI":"10.1126\/science.ade2574","type":"journal-article","article-title":"Evolutionary-scale prediction of atomic-level protein structure with a language model","volume":"379","author":"Lin","year":"2023","journal-title":"Science"},{"key":"bib21","doi-asserted-by":"publisher","first-page":"225","DOI":"10.1021\/acschembio.7b00996","type":"journal-article","article-title":"Determinants and prediction of esterase substrate promiscuity patterns","volume":"13","author":"Mart\u00ednez-Mart\u00ednez","year":"2018","journal-title":"ACS Chemical Biology"},{"key":"bib22","doi-asserted-by":"publisher","first-page":"1781","DOI":"10.1093\/molbev\/msh194","type":"journal-article","article-title":"Comparison of site-specific rate-inference methods for protein sequences: empirical Bayesian methods are superior","volume":"21","author":"Mayrose","year":"2004","journal-title":"Molecular Biology and Evolution"},{"key":"bib23","doi-asserted-by":"publisher","first-page":"2214","DOI":"10.1111\/febs.16274","type":"journal-article","article-title":"Enzyme nomenclature and classification: the state of the art","volume":"290","author":"McDonald","year":"2023","journal-title":"The FEBS Journal"},{"key":"bib24","doi-asserted-by":"publisher","first-page":"1145","DOI":"10.1101\/gr.277675.123","type":"journal-article","article-title":"Leveraging protein language models for accurate multiple sequence alignments","volume":"33","author":"McWhite","year":"2023","journal-title":"Genome Research"},{"key":"bib25","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2402.06748","type":"preprint","article-title":"CLIPZyme: reaction-conditioned virtual screening of enzymes","volume-title":"arXiv","author":"Mikhael","year":"2024"},{"key":"bib26","type":"software","article-title":"Squidly","volume-title":"Software Heritage","author":"Mora","year":"2026","unstructured":"Mora H, WRiegs, HerrLuca9. 2026. Squidly. Software Heritage. swh:1:rev:58a8f7d6cac128c4d0915835d20ecb575cb72931. https:\/\/archive.softwareheritage.org\/swh:1:dir:e001d14fa0439832623c66ec990d3d5a17e87eed;origin=https:\/\/github.com\/WRiegs\/Squidly;visit=swh:1:snp:ccd973dd1085ee375f1c7ffba0fa899f10bd1826;anchor=swh:1:rev:58a8f7d6cac128c4d0915835d20ecb575cb72931."},{"key":"bib27","doi-asserted-by":"publisher","DOI":"10.1093\/gpbjnl\/qzae094","type":"journal-article","article-title":"SCREEN: a graph-based contrastive learning tool to infer catalytic residues and assess enzyme mutations","volume":"1","author":"Pan","year":"2024","journal-title":"Genomics, Proteomics and Bioinformatics"},{"key":"bib28","doi-asserted-by":"publisher","DOI":"10.1186\/1471-2105-7-312","type":"journal-article","article-title":"Prediction of catalytic residues using support vector machine with selected protein sequence and structural properties","volume":"7","author":"Petrova","year":"2006","journal-title":"BMC Bioinformatics"},{"key":"bib29","doi-asserted-by":"publisher","first-page":"240","DOI":"10.1021\/acsbiomedchemau.2c00085","type":"journal-article","article-title":"Catalytic site proximity profiling for functional unification of sequence-diverse radical S -adenosylmethionine enzymes","volume":"3","author":"Precord","year":"2023","journal-title":"ACS Bio & Med Chem Au"},{"key":"bib30","doi-asserted-by":"publisher","DOI":"10.1101\/2020.12.15.422761","type":"preprint","article-title":"Transformer protein language models are unsupervised structure learners","volume-title":"bioRxiv","author":"Rao","year":"2020"},{"key":"bib31","doi-asserted-by":"publisher","DOI":"10.1016\/j.cbpa.2024.102536","type":"journal-article","article-title":"Catalyzing the future: recent advances in chemical synthesis using enzymes","volume":"83","author":"Reisenbauer","year":"2024","journal-title":"Current Opinion in Chemical Biology"},{"key":"bib32","doi-asserted-by":"publisher","first-page":"D618","DOI":"10.1093\/nar\/gkx1012","type":"journal-article","article-title":"Mechanism and catalytic site atlas (M-CSA): a database of enzyme reaction mechanisms and active sites","volume":"46","author":"Ribeiro","year":"2018","journal-title":"Nucleic Acids Research"},{"key":"bib33","doi-asserted-by":"publisher","first-page":"314","DOI":"10.1074\/jbc.REV119.006289","type":"journal-article","article-title":"A global analysis of function and conservation of catalytic residues in enzymes","volume":"295","author":"Ribeiro","year":"2020","journal-title":"The Journal of Biological Chemistry"},{"key":"bib34","doi-asserted-by":"publisher","first-page":"4277","DOI":"10.1021\/acs.jcim.3c00273","type":"journal-article","article-title":"A highly sensitive model based on graph neural networks for enzyme key catalytic residue prediction","volume":"63","author":"Shen","year":"2023","journal-title":"Journal of Chemical Information and Modeling"},{"key":"bib35","doi-asserted-by":"publisher","DOI":"10.1038\/msb.2011.75","type":"journal-article","article-title":"Fast, scalable generation of high-quality protein multiple sequence alignments using clustal omega","volume":"7","author":"Sievers","year":"2011","journal-title":"Molecular Systems Biology"},{"key":"bib36","doi-asserted-by":"publisher","first-page":"125","DOI":"10.1016\/j.jtbi.2018.01.023","type":"journal-article","article-title":"PREvaIL, an integrative approach for inferring catalytic residues using sequence, structural, and network features in a machine-learning framework","volume":"443","author":"Song","year":"2018","journal-title":"Journal of Theoretical Biology"},{"key":"bib37","doi-asserted-by":"publisher","first-page":"1026","DOI":"10.1038\/nbt.3988","type":"journal-article","article-title":"MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets","volume":"35","author":"Steinegger","year":"2017","journal-title":"Nature Biotechnology"},{"key":"bib38","doi-asserted-by":"publisher","DOI":"10.1093\/bib\/bbaf047","type":"journal-article","article-title":"Do protein language models learn phylogeny?","volume":"26","author":"Tule","year":"2024","journal-title":"Briefings in Bioinformatics"},{"key":"bib39","doi-asserted-by":"publisher","first-page":"243","DOI":"10.1038\/s41587-023-01773-0","type":"journal-article","article-title":"Fast and accurate protein structure search with Foldseek","volume":"42","author":"van Kempen","year":"2024","journal-title":"Nature Biotechnology"},{"key":"bib40","doi-asserted-by":"publisher","DOI":"10.1101\/2020.06.26.174417","type":"preprint","article-title":"BERTology Meets Biology: Interpreting Attention in Protein Language Models","volume-title":"arXiv","author":"Vig","year":"2021"},{"key":"bib41","doi-asserted-by":"publisher","DOI":"10.1186\/1471-2105-14-S3-S13","type":"journal-article","article-title":"Protein function annotation with structurally aligned local sites of activity (SALSAs)","volume":"14 Suppl 3","author":"Wang","year":"2013","journal-title":"BMC Bioinformatics"},{"key":"bib42","doi-asserted-by":"publisher","DOI":"10.1186\/s13321-024-00920-2","type":"journal-article","article-title":"Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning","volume":"16","author":"Wang","year":"2024","journal-title":"Journal of Cheminformatics"},{"key":"bib43","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-024-51511-6","type":"journal-article","article-title":"Multi-modal deep learning enables efficient and accurate annotation of enzymatic active sites","volume":"15","author":"Wang","year":"2024","journal-title":"Nature Communications"},{"key":"bib44","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2406.15669","type":"preprint","article-title":"CARE: a benchmark suite for the classification and retrieval of enzymes","volume-title":"arXiv","author":"Yang","year":"2025"},{"key":"bib45","doi-asserted-by":"publisher","first-page":"216","DOI":"10.1110\/ps.062523907","type":"journal-article","article-title":"Evaluation of features for catalytic residue prediction in novel folds","volume":"16","author":"Youn","year":"2007","journal-title":"Protein Science"},{"key":"bib46","doi-asserted-by":"publisher","DOI":"10.1016\/j.biotechadv.2022.107926","type":"journal-article","article-title":"Hot spots-making directed evolution easier","volume":"56","author":"Yu","year":"2022","journal-title":"Biotechnology Advances"},{"key":"bib47","doi-asserted-by":"publisher","first-page":"1358","DOI":"10.1126\/science.adf2465","type":"journal-article","article-title":"Enzyme function prediction using contrastive learning","volume":"379","author":"Yu","year":"2023","journal-title":"Science"},{"key":"bib48","doi-asserted-by":"publisher","first-page":"2329","DOI":"10.1093\/bioinformatics\/btn433","type":"journal-article","article-title":"Accurate sequence-based prediction of catalytic residues","volume":"24","author":"Zhang","year":"2008","journal-title":"Bioinformatics"},{"key":"dataset1","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.15541320","type":"dataset","article-title":"Models and datasets for Squidly","volume-title":"Zenodo","author":"Rieger","year":"2025","unstructured":"Rieger WJF. 2025. Models and datasets for Squidly. Zenodo."}],"container-title":["eLife"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/cdn.elifesciences.org\/articles\/108186\/elife-108186-v1.xml","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/cdn.elifesciences.org\/articles\/108186\/elife-108186-v1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/elifesciences.org\/articles\/108186","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,4]],"date-time":"2026-09-04T14:05:32Z","timestamp":1788530732000},"score":1,"resource":{"primary":{"URL":"https:\/\/elifesciences.org\/articles\/108186"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,4]]},"references-count":49,"alternative-id":["10.7554\/eLife.108186"],"URL":"https:\/\/doi.org\/10.7554\/elife.108186","archive":["CLOCKSS"],"relation":{"has-preprint":[{"id-type":"doi","id":"10.1101\/2025.06.13.659624","asserted-by":"subject"}],"is-same-as":[{"id-type":"doi","id":"10.7554\/eLife.108186.3","asserted-by":"subject"}],"is-version-of":[{"id-type":"doi","id":"10.7554\/eLife.108186.1","asserted-by":"subject"},{"id-type":"doi","id":"10.7554\/eLife.108186.2","asserted-by":"subject"}],"is-supplemented-by":[{"id-type":"doi","id":"10.5281\/zenodo.15541320","asserted-by":"subject"},{"id-type":"uri","id":"https:\/\/archive.softwareheritage.org\/swh:1:dir:e001d14fa0439832623c66ec990d3d5a17e87eed;origin=https:\/\/github.com\/WRiegs\/Squidly;visit=swh:1:snp:ccd973dd1085ee375f1c7ffba0fa899f10bd1826;anchor=swh:1:rev:58a8f7d6cac128c4d0915835d20ecb575cb72931","asserted-by":"subject"}]},"ISSN":["2050-084X"],"issn-type":[{"value":"2050-084X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,4]]},"assertion":[{"value":"single anonymised","name":"peer_review_transparency","label":"Peer review transparency","group":{"name":"peer_review_taxonomy"}},{"value":"other reviewer(s), editor","name":"peer_review_interaction","label":"Peer review interaction","group":{"name":"peer_review_taxonomy"}},{"value":"review summaries, review reports, author\/editor communication, reviewer identities reviewer opt in, editor identities","name":"peer_review_published","label":"Peer review published","group":{"name":"peer_review_taxonomy"}},{"value":"open (sign in with ORCID iD required)","name":"post_publication_commenting","label":"Post publication commenting","group":{"name":"post_publication_commenting"}}],"article-number":"RP108186"}}