Plant uORF-pep database: A multi-evidence integrative resource for upstream open reading frames and their encoded peptides in plants.

Upstream open reading frames (uORFs) in 5' untranslated regions regulate downstream translation, and many encode functional peptides. Despite their importance, no comprehensive plant uORF database with experimental validation exists. Here we present the Plant uORF-pep Database (https://plantuorf-pep.com), a multi-evidence resource encompassing 16.2M uORF records across 30 plant species. The database integrates four evidence layers: (1) Ribo-seq translation evidence from six species, including uORF-level validation of 5.3K Arabidopsis uORFs; (2) cross-species conservation analysis across 10 plants, identifying 85K genes with conserved uORFs; (3) machine learning-based translation probability prediction for 5M uORFs (Random Forest, 78.9% accuracy); and (4) discovery of 6.6K conserved but unannotated uORF peptides prioritized as experimental candidates. The web interface supports gene search, batch query, and data download. By bridging computational prediction with experimental and evolutionary evidence, Plant uORF-pep fills a critical gap for plant uORF research and functional peptide discovery.