DreaMS mass spectrometry foundation model weights

Provider
Roman Bushuiev, Tomas Pluskal et al. (Pluskal lab)
License
MIT (source)
Size
8 files
Stored at
Hugging Face (https://huggingface.co/roman-bushuiev/DreaMS)
Access
Free and public. No account with the data provider is needed.
Category
Proteomics
Source
https://www.nature.com/articles/s41587-025-02663-3

About DreaMS mass spectrometry foundation model weights

Pre-trained weights for DreaMS (Deep Representations Empowering the Annotation of Mass Spectra), the transformer from the Pluskal lab (Bushuiev et al., Nature Biotechnology 2025) trained self-supervised on GeMS, a large collection of unannotated tandem mass spectra mined from the MassIVE GNPS repository, to embed MS/MS spectra for molecular networking, spectral similarity and property prediction. The mounted repo holds ssl_model.ckpt (the self-supervised model), embedding_model.ckpt and embedding_model_massspecgym.ckpt (fine-tuned embedding models, about 1.2 GB each) and a TorchScript export for non-Python use. Load the checkpoints with the dreams package from the mount path and pair them with the GNPS libraries or Enveda-180 archives. MIT licence.

How to load DreaMS mass spectrometry foundation model weights in Python

import os, json
base = "/mnt/archive"   # this model repo, mounted read-only
files = sorted(os.listdir(base)); print(files)
print("DreaMS_embedding_model_torchscript.pt", os.path.getsize(os.path.join(base, "DreaMS_embedding_model_torchscript.pt")) // 2**20, "MB")

Written for a Scigantic notebook, where /mnt/archive is the mounted dataset.

Data structure

Files

Example file: DreaMS_embedding_model_torchscript.pt

Analyze this dataset without downloading it

On Scigantic, DreaMS mass spectrometry foundation model weights opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Proteomics datasets