{"id":1492,"date":"2026-07-07T11:41:28","date_gmt":"2026-07-07T09:41:28","guid":{"rendered":"https:\/\/bdagroup.nl\/?post_type=internship&#038;p=1492"},"modified":"2026-07-07T11:55:57","modified_gmt":"2026-07-07T09:55:57","slug":"designing-terpene-synthases-with-generative-ai-protein-models","status":"publish","type":"internship","link":"https:\/\/bdagroup.nl\/?internship=designing-terpene-synthases-with-generative-ai-protein-models","title":{"rendered":"Designing Terpene Synthases with Generative AI Protein Models"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">Terpene synthases (TPSs) produce terpenes, useful molecules for fragrances, flavors, and medicine. Designing new TPS by hand is hard: minor active-site changes can completely change the product, and there is little usable sequence\u2013function relationship to guide edits. Generative AI is starting to change this. A recent example,<a href=\"https:\/\/consensus.app\/papers\/details\/e14399fb79695890a0ca6b59516a43bb\/?utm_source=claude_desktop\"> <\/a>TpsGPT <a href=\"https:\/\/doi.org\/10.48550\/arXiv.2512.08772\" target=\"_blank\" rel=\"noreferrer noopener\">(Ramanathan et al., 2025)<\/a>, fine-tuned a protein language model (ProtGPT2 <a href=\"https:\/\/doi.org\/10.1038\/s41467-022-32007-7\" target=\"_blank\" rel=\"noreferrer noopener\">(Ferruz et al., 2022)<\/a>) on ~79,000 TPS sequences from UniProt <a href=\"https:\/\/doi.org\/10.1093\/nar\/gkae1010\" target=\"_blank\" rel=\"noreferrer noopener\">(\u201cUniProt,\u201d 2024)<\/a> and generated tens of thousands of de novo candidates; after strict filtering, seven passed all checks and at least two were experimentally confirmed to be active enzymes.\u00a0<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">This project builds directly on that result but takes a different route: it works from protein structure rather than sequence. AI models such as AlphaFold <a href=\"https:\/\/doi.org\/10.1038\/s41586-021-03819-2\" target=\"_blank\" rel=\"noreferrer noopener\">(Jumper et al., 2021)<\/a> and ESMFold <a href=\"https:\/\/doi.org\/10.1126\/science.ade2574\" target=\"_blank\" rel=\"noreferrer noopener\">(Lin et al., 2023)<\/a> can predict a 3D structure for any TPS sequence, and this structural information encodes the active-site geometry that actually governs the carbocation cyclization cascade determining product outcome. Based on an already assembled dataset of predicted TPS structures (unpublished) you will create structure-aware generative models, for example through structure-conditioned approaches that generate sequences fitted to a target fold or active-site shape.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Concretely, you will (I) train or fine-tune a generative model on structures, building on approaches such as RFdiffusion <a href=\"https:\/\/doi.org\/10.1038\/s41586-023-06415-8\" target=\"_blank\" rel=\"noreferrer noopener\">(Watson et al., 2023)<\/a> and Chroma <a href=\"https:\/\/doi.org\/10.1038\/s41586-023-06728-8\" target=\"_blank\" rel=\"noreferrer noopener\">(Ingraham et al., 2023)<\/a>; (II) generate variants and screen them with computational metrics, e.g. fold confidence, sequence diversity, domain detection, and structural alignment (Foldseek <a href=\"https:\/\/doi.org\/10.1038\/s41587-023-01773-0\" target=\"_blank\" rel=\"noreferrer noopener\">(van Kempen et al., 2024)<\/a>); and (III) shortlist the most promising designs for experimental validation.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\"><\/h2>\n","protected":false},"template":"","categories":[],"tags":[],"class_list":["post-1492","internship","type-internship","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/bdagroup.nl\/index.php?rest_route=\/wp\/v2\/internship\/1492","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/bdagroup.nl\/index.php?rest_route=\/wp\/v2\/internship"}],"about":[{"href":"https:\/\/bdagroup.nl\/index.php?rest_route=\/wp\/v2\/types\/internship"}],"wp:attachment":[{"href":"https:\/\/bdagroup.nl\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1492"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/bdagroup.nl\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1492"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/bdagroup.nl\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1492"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}