DDA-BERT is an end-to-end rescoring tool tailored for data-dependent acquisition (DDA) proteomics. Leveraging a deep learning model based on the Transformer architecture, it refines initially identified peptide-spectrum matches (PSMs) to improve identification accuracy and sensitivity. Trained on a large-scale dataset of 3,701 DDA-MS files and approximately 82 million high-confidence PSMs, DDA-BERT effectively captures complex relationships between peptide sequences and MS/MS spectra. The model delivers robust and consistent performance across a wide range of biological systems, including animal, plant, and microbial proteomes. It also demonstrates high sensitivity in low-input conditions, such as trace-level and single-cell proteomics, making it well-suited for diverse experimental contexts.

By leveraging an end-to-end architecture, DDA-BERT eliminates the need for search engine–specific feature engineering and can be readily integrated into diverse proteomics workflows, leading to improved sensitivity and accuracy in PSM analysis.

The software is actively maintained and continuously updated. We sincerely invite you to try it and welcome your feedback or suggestions at: ajun@westlake.edu.cn; guotiannan@westlake.edu.cn.