MassNet

training · DDA · 2 versions · 2 papers

AI-ready corpus of more than 1.5 billion DDA MS/MS spectra from 27,643 LC-MS runs in 648 public projects over 35 species, searched with FragPipe and Sage and stored in the MSDT tensor format.
Kind training
Acquisition DDA
Organisms 35 animal, plant and microbial species
Home https://guomics.com/software/massnet

AI-ready corpus of more than 1.5 billion DDA MS/MS spectra from 27,643 LC-MS runs in 648 public projects over 35 species, searched with FragPipe and Sage and stored in the MSDT tensor format.

Versions

full corpus

Raw MSDT files for every run, a metadata file, and global FDR-filtered FragPipe and Sage results. Deposited as PRIDE PXD071590, which was still private on 2026-10-15: the Nature Methods data statement gives reviewer access only. Assembled from 647 public projects (601 PRIDE, 46 iProX), listed in the paper’s Supplementary Table 1 and recorded here as provenance, plus 3,379 in-house raw files (Mus musculus, Homo sapiens) with no accession of their own.

introduced by MassNet: billion-scale AI-friendly mass spectral corpus enables robust de novo peptide sequencing

Where it lives:

Assembled from 647 third-party submissions:

XuanjiNovo training set (100M PSMs)

100 million PSMs sampled at random from the FragPipe identifications (Sage identifications not used), the training set of XuanjiNovo-100M. Part of PXD071590. The 30M and 65M scaling subsets are not published separately.

introduced by MassNet: billion-scale AI-friendly mass spectral corpus enables robust de novo peptide sequencing

Where it lives:

  • PRIDE · PXD071590 training set of 100 million PSMs

Deposited by (2)

More than one paper here means one piece of work published twice, a preprint and its version of record; a deposit itself happens once.

Trained on by (2)

Methods on these papers (1)

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.

Seen in the charts

Back to the full map

Back to top