MassIVE-KB
training · DDA · 3 versions · 38 papers
A repository-wide spectral library of human peptide PSMs, large enough to train on and cited by version number. No single accession stands for it, which is why papers name it rather than an identifier.
| Kind | training |
| Acquisition | DDA |
| Organisms | Homo sapiens |
| Home | https://massive.ucsd.edu/ProteoSAFe/static/massive-kb-libraries.jsp |
A repository-wide spectral library of human peptide PSMs, large enough to train on and cited by version number. No single accession stands for it, which is why papers name it rather than an identifier.
Versions
v1
The version Casanovo’s original training set was drawn from. MassIVE titles MSV000081142 “MassIVE-KB Knowledge Base spectral library of Human HCD spectra” and names no version itself, so attaching it here rests on it being the original human HCD library rather than on the record saying so.
Where it lives:
- MassIVE · MSV000081142
v2
The later, larger release; papers that say only “MassIVE-KB” are ambiguous between the two.
No public address: this version is named in the literature but cannot be downloaded.
v1 30M-PSM split (Casanovo)
Up to the top 100 PSMs for each of the 2,154,269 precursors in the MassIVE-KB v1 spectral library, split peptide-disjoint into training, validation and test sets; deposited by Wout Bittremieux as the data Casanovo is trained on.
spectra 30,504,897 · train 28,508,636 · validation 1,000,234 · test 996,027 · released 2025-03-10
Where it lives:
- DOI · 10.5281/zenodo.14967861
Trained on by (37)
- NovoCRF: Globally Normalized De Novo Peptide Sequencing on a Prefix-Mass Lattice (2026) preprint version not stated
- PepPr: De Novo peptide sequencing with peptide priors (2026) preprint v1 30M-PSM split (Casanovo)
- MassNet: billion-scale AI-ready mass spectral corpus empowering scalable deep learning in proteomics (2026) peer-reviewed version not stated
- dIon: Fragmentation-Based Invariance for Self-Supervised Learning of Tandem Mass Spectra (2026) preprint v2
- False discovery rate control for trustworthy AI-based de novo peptide sequencing (2026) preprint v1
- π-HelixNovo2: Making Accurate Online De Novo Peptide Sequencing Available to All (2026) peer-reviewed v1, v1 30M-PSM split (Casanovo)
- Zero-shot de novo peptide sequencing with open posttranslational modification discovery (2026) peer-reviewed version not stated
- XA-Novo: an accurate and high-throughput mass spectrometry-based de novo sequencing technology for monoclonal antibodies and antibody mixtures (2026) peer-reviewed v1
- Accurate de novo sequencing of the modified proteome with OmniNovo (2025) preprint v1, v2
- Bidirectional Representations Augmented Autoregressive Biological Sequence Generation (2025) ML conference v1 30M-PSM split (Casanovo)
- Bidirectional Representations Augmented Autoregressive Biological Sequence Generation (2025) preprint v1 30M-PSM split (Casanovo)
- Modanovo: A Unified Model for Post-Translational Modification-Aware de Novo Sequencing Using Experimental Spectra from In Vivo and Synthetic Peptides (2025) preprint v1 30M-PSM split (Casanovo)
- Improvements to CasaNovo, a deep learning de novo peptide sequencer (2025) preprint v1, v2
- Curriculum Learning for Biological Sequence Prediction: The Case of De Novo Peptide Sequencing (2025) ML conference v1 30M-PSM split (Casanovo)
- Universal Biological Sequence Reranking for Improved De Novo Peptide Sequencing (2025) ML conference v1 30M-PSM split (Casanovo)
- A transformer model for de novo sequencing of data independent acquisition mass spectrometry data (2025) peer-reviewed v1 30M-PSM split (Casanovo)
- Zero-Shot De Novo Peptide Sequencing with Open Post-Translational Modification Discovery (2025) preprint version not stated
- MassNet: billion-scale AI-friendly mass spectral corpus enables robust de novo peptide sequencing (2025) preprint version not stated
- Curriculum Learning for Biological Sequence Prediction: The Case of De Novo Peptide Sequencing (2025) preprint v1 30M-PSM split (Casanovo)
- Universal Biological Sequence Reranking for Improved De Novo Peptide Sequencing (2025) preprint v1 30M-PSM split (Casanovo)
- Pairwise Attention: Leveraging Mass Differences to Enhance De Novo Sequencing of Mass Spectra (2025) preprint version not stated
- Distilling Non-Autoregressive Model Knowledge for Autoregressive De Novo Peptide Sequencing (2025) preprint v1 30M-PSM split (Casanovo)
- π-PrimeNovo: an accurate and efficient non-autoregressive deep learning model for de novo peptide sequencing (2025) peer-reviewed v1
- RankNovo: A Universal Reranking Approach for Robust De Novo Peptide Sequencing (2024) preprint v1 30M-PSM split (Casanovo)
- BERT model for de novo protein sequencing using tandem mass spectra (2024) abstract version not stated
- Accounting for Digestion Enzyme Bias in Casanovo (2024) peer-reviewed v1 30M-PSM split (Casanovo)
- Transforming de novo peptide sequencing by explainable AI (2024) preprint v1
- Sequence-to-sequence translation from mass spectra to peptides with a transformer model (2024) peer-reviewed v1 30M-PSM split (Casanovo)
- PowerNovo: de novo peptide sequencing via tandem mass spectrometry using an ensemble of transformer and BERT models (2024) peer-reviewed v1
- A transformer model for de novo sequencing of data independent acquisition mass spectrometry data (2024) preprint v1 30M-PSM split (Casanovo)
- π-PrimeNovo: an accurate and efficient non-autoregressive deep learning model for de novo peptide sequencing (2024) preprint v1
- ContraNovo: A Contrastive Learning Approach to Enhance De Novo Peptide Sequencing (2024) peer-reviewed v1 30M-PSM split (Casanovo)
- Introducing π-HelixNovo for practical large-scale de novo peptide sequencing (2024) peer-reviewed v1
- Accurate de novo peptide sequencing using fully convolutional neural networks (2023) peer-reviewed version not stated
- Introducing PandaNovo for practical large-scale de novo peptide sequencing (2023) preprint v1
- Sequence-to-sequence translation from mass spectra to peptides with a transformer model (2023) preprint v1 30M-PSM split (Casanovo)
- PepNet: A Fully Convolutional Neural Network for De novo Peptide Sequencing (2022) preprint version not stated
Evaluated on by (5)
- Protein Language Model-Aligned Spectra Embeddings for De Novo Peptide Sequencing (2025) preprint version not stated
- Modanovo: A Unified Model for Post-Translational Modification-Aware de Novo Sequencing Using Experimental Spectra from In Vivo and Synthetic Peptides (2025) preprint v1 30M-PSM split (Casanovo), v2
- Improvements to CasaNovo, a deep learning de novo peptide sequencer (2025) preprint v1, v2
- Accounting for Digestion Enzyme Bias in Casanovo (2024) peer-reviewed v1 30M-PSM split (Casanovo)
- PowerNovo: de novo peptide sequencing via tandem mass spectrometry using an ensemble of transformer and BERT models (2024) peer-reviewed v1
Checkpoints trained on this (6)
| Method | Checkpoint | Version of this dataset | Size | Get it |
|---|---|---|---|---|
| Casanovo | v4.0.0 | v1 | — | GitHub release |
| Casanovo | v4.2.0 | v1 | — | GitHub release |
| Casanovo | v4.2.0 | v2 | — | GitHub release |
| RefineNovo | RefineNovo-30M | not stated | 388 MB | Google Drive · copy |
| π-PrimeNovo | MassIVE | not stated | 387 MB | Google Drive · copy |
| π-PrimeNovo | phosphorylation | not stated | 387 MB | Google Drive · copy |
Each link rests on stated evidence rather than a text match:
- Casanovo v4.0.0: trained_on: ‘MassIVE-KB v1’, from ProteoBench discussion 356
- Casanovo v4.2.0: trained_on: ‘~2M PSMs from MassIVE-KB v1 + v2.0.15’
- Casanovo v4.2.0: trained_on: ‘~2M PSMs from MassIVE-KB v1 + v2.0.15’
- RefineNovo RefineNovo-30M: the FILENAME, model_massivekb.ckpt; BEAM-Labs/denovo states no training set for it
- π-PrimeNovo MassIVE: trained_on: ‘MassIVE-KB’, with no version given
- π-PrimeNovo phosphorylation: INHERITED, not direct: fine-tuned from model_massive.ckpt, which was trained on MassIVE-KB. Its own fine-tuning set, 2020-Cell-LUAD, is not in this catalog
Methods on these papers (24)
- ByNovo algorithm
- Casanovo algorithm
- Cascadia algorithm
- ContraNovo algorithm
- CrossNovo algorithm
- FDR control for AI-based de novo sequencing post-processor
- Modanovo algorithm
- NovoCRF algorithm
- OmniNovo algorithm
- PLMNovo algorithm
- Pairwise algorithm
- PepNet algorithm
- PepPr algorithm
- PowerNovo algorithm
- RNovA algorithm
- RankNovo post-processor
- RefineNovo algorithm
- XA-Novo algorithm
- XuanjiNovo algorithm
- dIon algorithm
- π-HelixNovo algorithm
- π-HelixNovo2 algorithm
- π-PrimeNovo algorithm
- π-xNovo algorithm
Taken from the describing links only, so a paper that merely ran a tool on this data does not make that tool a method of it.