ArticlePublished: 18 April 2026Tengda Li ORCID: orcid.org/0009-0006-9984-80961,2 &Xiang Li ORCID: orcid.org/0009-0007-5694-69531,3 Oncogene (2026)Cite this articleSubjectsGene expression profilingProteomicsSequencingAbstractProtein S-palmitoylation, a dynamic lipid modification, is essential for protein stability, trafficking, and signaling; dysregulated palmitoyltransferases drive cancer, yet systematic discovery of palmitoyltransferases remains hindered by labor-intensive, motif-dependent assays. We present iPalmT, an end-to-end deep learning framework that identifies palmitoyltransferases directly from primary amino acid sequence without handcrafted features or prior domain annotations. The model combines convolutional layers and squeeze-and-excitation mechanisms to capture local sequence signals and long-range dependencies. On an independent test set, iPalmT achieved 0.99 accuracy, 0.98 precision, 0.97 recall, and 0.98 F1 score. Integrated Gradients attribution emphasized the canonical DHHC motif and highlighted additional putative functional domains, despite receiving no motif supervision. Proteome-scale application to human sequences yielded unreviewed candidates; two (A0A0D9SEX5 and A0A1W2PRJ8) underwent structural analysis and experimental validation, which supported the predictions. We further release a large predicted palmitoyltransferase resource comprising 10,365,644 sequences identified from 147,847,003 proteins across 33,285 species-level groups to support large-scale exploration and cross-species analyses. iPalmT is available as a standalone program (https://github.com/Tengda-Li-Lab/iPalmT.git), offering a scalable, sequence-only route to discover noncanonical, evolutionarily divergent palmitoyltransferases.This is a preview of subscription content, access via your institutionAccess optionsSubscribe to this journalReceive 50 print issues and online access269,00 € per yearonly 5,38 € per issueLearn moreBuy this articlePurchase on SpringerLinkInstant access to the full article PDF.39,95 €Prices may be subject to local taxes which are calculated during checkoutFig. 1: Network architecture of iPalmT.The alternative text for this image may have been generated using AI.Fig. 2: Evaluation of the identification efficiency of iPalmT and comparison with other models.The alternative text for this image may have been generated using AI.Fig. 3: Domains that were highlighted in the identification process of the iPalmT model.The alternative text for this image may have been generated using AI.Fig. 4: Analysis of the structures and sequences of proteins identified as palmitoyltransferases by the iPalmT model but annotated as unreviewed in Uniprot.The alternative text for this image may have been generated using AI.Fig. 5: Validation of the palmitoyltransferase activity of A0A0D9SEX5 and A0A1W2PRJ8.The alternative text for this image may have been generated using AI.Data availabilityThe large-scale compendium of iPalmT predictions for candidate palmitoyltransferases is hosted at https://doi.org/10.5281/zenodo.17159856. Summary tables for the two palmitoylation proteomics results generated in this study are presented in Tables S7 and S8. Other data will be made available on request.Code availabilityThe original iPalmT source code is openly available on GitHub (https://github.com/Tengda-Li-Lab/iPalmT.git).ReferencesS Mesquita F, Abrami L, Linder ME, Bamji SX, Dickinson BC, van der Goot FG. Mechanisms and functions of protein S-acylation. Nat Rev Mol Cell Biol. 2024;25:488–509.Article CAS PubMed PubMed Central Google Scholar Yang X, Xu M, Deng Z, Xu B. The shadow of cancer therapeutic resistance: Unveiling the role of S-palmitoylation. Drug Resist Updates. 2025;82:101264.Article CAS Google Scholar Blanc M, David F, Abrami L, Migliozzi D, Armand F, Bürgi J, et al. SwissPalm: Protein Palmitoylation database. F1000Research. 2015;4:261.Article PubMed PubMed Central Google Scholar Ren J, Wen L, Gao X, Jin C, Xue Y, Yao X. CSS-Palm 2.0: an updated software for palmitoylation sites prediction. Protein Eng Des Sel. 2008;21:639–44.Article CAS PubMed PubMed Central Google Scholar Kumari B, Kumar R, Kumar M. PalmPred: an SVM based palmitoylation prediction method using sequence profile information. PLoS ONE. 2014;9:e89246.Article PubMed PubMed Central Google Scholar Mistry J, Chuguransky S, Williams L, Qureshi M, Salazar Gustavo A, Sonnhammer ELL, et al. Pfam: the protein families database in 2021. Nucleic Acids Res. 2020;49:D412–9.Article Google Scholar Blum M, Chang H-Y, Chuguransky S, Grego T, Kandasaamy S, Mitchell A, et al. The InterPro protein families and domains database: 20 years on. Nucleic Acids Res. 2020;49:D344–54.Article Google Scholar Eddy SR. Profile hidden Markov models. Bioinformatics. 1998;14:755–63.Article CAS PubMed Google Scholar Eddy SR. Accelerated Profile HMM Searches. PLOS Comput Biol. 2011;7:e1002195.Article CAS PubMed PubMed Central Google Scholar Boadu F, Lee A, Cheng J. Deep learning methods for protein function prediction. Proteomics. 2025;25:e2300471.Article PubMed Google Scholar Chandra A, Tünnermann L, Löfstedt T, Gratz R. Transformer-based deep learning for predicting protein properties in the life sciences. eLife. 2023;12:e82819.Article CAS PubMed PubMed Central Google Scholar Kim GB, Gao Y, Palsson BO, Lee SY. DeepTFactor: a deep learning-based tool for the prediction of transcription factors. Proc Natl Acad Sci USA. 2021;118:e2021171118.Pakhrin SC, Pokharel S, Saigo H, Kc DB. Deep learning-based advances in protein posttranslational modification site and protein cleavage prediction. Methods Mol Biol. 2022;2499:285–322.Article PubMed Google Scholar Wang W, Shuai Y, Zeng M, Fan W, Li M. DPFunc: accurately predicting protein function via deep learning with domain-guided structure information. Nat Commun. 2025;16:70.Article CAS PubMed PubMed Central Google Scholar Li T. Method and system for predicting protein palmitoyltransferases based on multi-branch deep convolutional neural networks. Chinese Patent CN 121122422B. 2026.Krogh A, Larsson B, von Heijne G, Sonnhammer EL. Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J Mol Biol. 2001;305:567–80.Article CAS PubMed Google Scholar Sonnhammer EL, von Heijne G, Krogh A. A hidden Markov model for predicting transmembrane helices in protein sequences. In: Proceedings of the international conference on intelligent systems for molecular biology. AAAI Press, Menlo Park, CA. 1998;6:175–82.Nakamura MT, Yudell BE, Loor JJ. Regulation of energy metabolism by long-chain fatty acids. Prog Lipid Res. 2014;53:124–44.Article CAS PubMed Google Scholar Choi RY, Coyner AS, Kalpathy-Cramer J, Chiang MF, Campbell JP. Introduction to machine learning, neural networks, and deep learning. Transl Vis Sci Technol. 2020;9:14.PubMed PubMed Central Google Scholar van der Velden BHM, Kuijf HJ, Gilhuijs KGA, Viergever MA. Explainable artificial intelligence (XAI) in deep learning-based medical image analysis. Med Image Anal. 2022;79:102470.Article PubMed Google Scholar Lin Z, Akin H, Rao R, Hie B, Zhu Z, Lu W, et al. Evolutionary-scale prediction of atomic-level protein structure with a language model. Science. 2023;379:1123–30.Article CAS PubMed Google Scholar Muhammad D, Bendechache M. Unveiling the black box: A systematic review of Explainable Artificial Intelligence in medical image analysis. Comput Struct Biotechnol J. 2024;24:542–60.Article PubMed PubMed Central Google Scholar Talukder A, Barham C, Li X, Hu H. Interpretation of deep learning in genomics and epigenomics. Brief Bioinform. 2021;22:bbaa177.Nensa F, Demircioglu A, Rischpler C. Artificial intelligence in nuclear medicine. J Nucl Med. 2019;60:29s–37s.Article PubMed Google Scholar Oh JH, Kim HG, Lee KM. Developing and evaluating deep learning algorithms for object detection: key points for achieving superior model performance. Korean J Radiol. 2023;24:698–714.Article PubMed PubMed Central Google Scholar Dauparas J, Anishchenko I, Bennett N, Bai H, Ragotte RJ, Milles LF, et al. Robust deep learning-based protein sequence design using ProteinMPNN. Science. 2022;378:49–56.Article CAS PubMed PubMed Central Google Scholar Gainza P, Sverrisson F, Monti F, Rodolà E, Boscaini D, Bronstein MM, et al. Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning. Nat Methods. 2020;17:184–92.Article CAS PubMed Google Scholar Wang J, Lisanza S, Juergens D, Tischer D, Watson JL, Castro KM, et al. Scaffolding protein functional sites using deep learning. Science. 2022;377:387–94.Article CAS PubMed PubMed Central Google Scholar Baisya DR, Lonardi S. Prediction of histone post-translational modifications using deep learning. Bioinformatics. 2021;36:5610–7.Article PubMed Google Scholar Lipkova J, Chen RJ, Chen B, Lu MY, Barbieri M, Shao D, et al. Artificial intelligence for multimodal data integration in oncology. Cancer Cell. 2022;40:1095–110.Article CAS PubMed PubMed Central Google Scholar Shen P, Yang Z, Sun J, Wang Y, Qiu C, Wang Y, et al. Explainable multimodal deep learning for predicting thyroid cancer lateral lymph node metastasis using ultrasound imaging. Nat Commun. 2025;16:7052.Article CAS PubMed PubMed Central Google Scholar Download referencesAcknowledgementsThe computations in this research were performed using the CFFF platform of Fudan University.FundingThis work was supported by the National Natural Science Foundation of China (grant number 82503998) and the Talent Research Startup Project of Wenzhou Medical University (grant number QTJ21020).Author informationAuthors and AffiliationsZhejiang Provincial Key Laboratory of Medical Genetics, Key Laboratory of Laboratory Medicine, Ministry of Education, School of Laboratory Medicine and Life Sciences, Wenzhou Medical University Wenzhou, Zhejiang, ChinaTengda Li & Xiang LiDepartment of Chemistry, Fudan University, Shanghai, ChinaTengda LiCixi Biomedical Research Institute, Wenzhou Medical University, Zhejiang, ChinaXiang LiAuthorsTengda LiView author publicationsSearch author on:PubMed Google ScholarXiang LiView author publicationsSearch author on:PubMed Google ScholarContributionsTengda Li was responsible for code development, model optimization, sample collection and testing, as well as the overall experimental design, manuscript conceptualization, writing, and revision. Xiang Li made a contribution to the experiments. The final version of the manuscript has been reviewed and approved by all authors.Corresponding authorCorrespondence to Tengda Li.Ethics declarationsCompeting interestsThe authors declare no competing interests.Additional informationPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary informationTable S1. Summary of procedures for positive-sample verification, redundancy removal, and homology-leakage prevention. (download XLSX )41388_2026_3802_MOESM2_ESM.xlsx (download XLSX )Table S2. Summary of the retrieved UniProt entries initially designated as positive candidates (n = 224,455, Palmitoyltransferase-related items were highlighted in blue).Table S3. Optimization and Selection of Convolutional Neural Network Architectures (download XLSX )Table S4. Determining Optimal Hyperparameters Based on Prediction Accuracy (download XLSX )Table S5. Quantitative Comparative Evaluation of iPalmT Against Other Models(CI, confidence intervals) (download XLSX )Table S6. Key Structural Regions Comprising High-Scoring Residues Identified by iPalmT (download XLSX )Table S7. Results of palmitoyl-proteomic profiling in Urotsa cells (A0A0D9SEX5 vs. Vector). (download XLSX )Table S8. Results of palmitoyl-proteomic profiling in Urotsa cells (A0A1W2PRJ8 vs. Vector). (download XLSX )Figures S1–S4 (download PDF )Rights and permissionsSpringer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.Reprints and permissionsAbout this article