A large-scale unified deep learning model for peptide mass spectrum interpretation trained on multimodal data
peer-reviewed · Nature Machine Intelligence · 2026
peer-reviewed · Nature Machine Intelligence · 2026. Jiale Zhao et al.
| Date | 2026-05-25 |
| Type | peer-reviewed |
| Venue | Nature Machine Intelligence |
| Publisher | Springer Science and Business Media LLC |
| Contribution | algorithm |
| DOI | 10.1038/s42256-026-01234-8 |
| Citations (OpenAlex) | 1 |
| Venue 2-year citedness | 23.00 |
Preprint version: pUniFind: a unified large pre-trained deep learning model pushing the limit of mass spectra interpretation (2025-06-30, arXiv)
Methods and tools
- pUniFind: Multimodal pre-trained transformer for mass spectra that unifies peptide-spectrum scoring and zero-shot de novo sequencing in a single model. Trained on >100M open-search-derived spectra; reports +60% PSMs over prior de novo methods with 1,300+ modifications supported, and a DL-based QC step that recovers 38.5% additional peptides.
Data used
- A draft map of the human proteome (as deposited) · PXD000561
- A mass-tolerant database search identifies a large proportion of unassigned spectra in shotgun proteomics as modified pe (as deposited) · PXD001468
- Cell-type and brain-region resolved mouse brain proteome (as deposited) · PXD001250
- Class II Immunopeptidome of 9033 cells (as deposited) · PXD029648
- Confetti: A Multi-protease Map of the HeLa Proteome for Comprehensive Proteomics (as deposited) · PXD000900
- HLA-DQ8 Immunopeptidomics, Type 1 Diabetes (as deposited) · PXD019466
- HLA-I peptidomics od Meningioma tissues - Peptide length distribution and multiple specificity in naturally presented HL (as deposited) · PXD009925
- HeLa proteome of 12,250 protein-coding genes (as deposited) · PXD004452
- Human Testis LC-MS/MS - Multi-Protease Strategy Identifies Three PE2 Missing Proteins in Human Testis Tissue (as deposited) · PXD006465
- Human testis off-line LC-MS/MS (PXD009737) (as deposited) · PXD009737
- IPX000540500038 (as deposited) · IPX000540500038
- Machine learning predictions of MHC-II specificities reveal alternative binding mode of class II epitopes (as deposited) · PXD034773
- Mass spectrometry based draft of the human proteome (as deposited) · PXD000865
- MaxQuant software for ion mobility enhanced shotgun proteomics (as deposited) · PXD014777
- Minimalistic sample processing (as deposited) · PXD000269
- Modulating the selectivity of affinity absorbents to multi-phosphopeptides by a novel competitive substitution strategy (as deposited) · PXD004252
- Multi-omics profiling of human pancreatic islet dysregulation from normoglycemia to type 2 diabetes (as deposited) · PXD022561
- Multienzyme deep learning models improve peptide de novo sequencing by mass spectrometry proteomics (as deposited) · PXD037803
- Online parallel accumulation – serial fragmentation (PASEF) with a novel trapped ion mobility mass spectrometer (as deposited) · PXD010012
- ProteomeTools (Parts I-III) · PXD004732, PXD010595, PXD021013
- Reproducibility of label-free single-shot phosphoproteomics applied to CRC cell lines (as deposited) · PXD001546
- Synthetic (Phospho)Peptide Library (as deposited) · PXD000138
- The HLA-Ligand-Atlas. A resource of natural HLA ligands presented on benign tissues (as deposited) · PXD019643
- The Proteome Landscape of the Kingdoms of Life (as deposited) · PXD014877, PXD019483
- The immunopeptidomic landscape of ovarian carcinoma (as deposited) · PXD007635
- hnRNPR interactome in axons and soma of murine motoneurons (as deposited) · PXD043851
Cites (14)
- π-PrimeNovo: an accurate and efficient non-autoregressive deep learning model for de novo peptide sequencing (2025) crossref
- NovoBoard: A Comprehensive Framework for Evaluating the False Discovery Rate and Accuracy of De Novo Peptide Sequencing (2024) crossref
- A Handle on Mass Coincidence Errors in De Novo Sequencing of Antibodies by Bottom-up Proteomics (2024) crossref
- Sequence-to-sequence translation from mass spectra to peptides with a transformer model (2024) crossref
- Mitigating the missing-fragmentation problem in de novo peptide sequencing with a two-stage graph-based deep learning model (2023) crossref
- Multienzyme deep learning models improve peptide de novo sequencing by mass spectrometry proteomics (2023) crossref
- De novo mass spectrometry peptide sequencing with a transformer model (2022) crossref
- Computationally instrument-resolution-independent de novo peptide sequencing for high-resolution devices (2021) crossref
- pNovo 3: precise de novo peptide sequencing using a learning-to-rank framework (2019) crossref
- Prosit: proteome-wide prediction of peptide tandem mass spectra by deep learning (2019) crossref
- De novo peptide sequencing by deep learning (2017) crossref
- Open-pNovo: De Novo Peptide Sequencing with Thousands of Protein Modifications (2017) crossref
- PepNovo: de novo peptide sequencing via probabilistic network modeling (2005) crossref
- PEAKS: powerful software for peptide de novo sequencing by tandem mass spectrometry (2003) crossref