DFlowNovo De Novo Peptide Sequencing

DFlowNovo De Novo Peptide Sequencing

Predicts peptide sequences from tandem mass spectrometry files with a flow matching model, returning calculated masses and confidence scores, with results you can save as CSV or TSV. It suits proteomics researchers.

Made by joelinator on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: DFlowNovo is a Gradio app that joelinator published as a Hugging Face Space. It reads tandem mass spectrometry (MS/MS) spectrum files and predicts the peptide sequences behind them, with calculated masses and confidence scores. It uses flow matching on discrete states, built on continuous-time Markov chains, to decode the amino acids quickly and in parallel.

Who it suits: Proteomics researchers and students who work with MS/MS spectra, and bioinformatics developers who want to read or run a model-based sequencing demo.

Industry | Who uses it | How they use it | Example
Proteomics | Mass spectrometry researchers | Upload an .mgf or .mgz spectrum file and run the sequencing | Getting peptide predictions with masses and confidence scores for their spectra
Bioinformatics | Method developers | Run the same file through different model checkpoints | Comparing the production and PTM extended checkpoints on one file
Education | Students | Try the built-in sample spectrum | Clicking the example and reading the prediction table

Skip it if you need a general protein tool, since it takes mass spectrometry files and nothing else.

Where to use it: DFlowNovo runs in these places.

• A web browser: open the Space's page on Hugging Face
• Python on Windows, macOS or Linux: run a copy of the Space

Use case | Who | What they get back
Sequence a spectrum file | Proteomics researchers | Peptide predictions with calculated masses and confidence scores
Run a quick test | Newcomers | Results from the built-in sample spectrum, with no file of your own
Export the results | Analysts | A CSV or TSV file of the predictions

How to set it up: Online use needs only a browser. A local copy needs a few things first.

• A web browser for the online Space
• Python on Windows, macOS or Linux for a local copy
• The Python packages listed in requirements.txt

1. Install the Hugging Face hub library with pip.
2. Download the Space's files into a folder called dflow-novo.
3. Move into that folder.
4. Install Gradio at the version the Space uses.
5. Install the packages listed in requirements.txt.
6. Start the app with Python.

pip install -U huggingface_hub
hf download joelinator/dflow-novo --repo-type space --local-dir dflow-novo
cd dflow-novo
pip install gradio==5.50.0
pip install -r requirements.txt
python app.py

How to use it: Online, open the Space's page and upload a spectrum file in .mgz, .mgf, .gz or .txt format, or use the built-in sample. Pick a model checkpoint, click Run De Novo Sequencing, then download the results. On your own computer, start the app from the folder with the command below.

python app.py

Tool | What it does
Upload Spectrum File | Takes a mass spectrometry file in .mgz, .mgf, .gz or .txt format
Clickable Example (.mgz) | Loads the built-in sample spectrum for a quick test
Select Model Checkpoint | Picks one of three checkpoints: production, MassIVE-KB pretrained or PTM extended
Run De Novo Sequencing | Generates the peptide predictions, calculated masses and confidence scores
Download | Saves the results as a CSV or TSV file
python app.py | Starts the Gradio app from the main file

Example: You open the Space, click the built-in sample spectrum, choose the production checkpoint and click Run De Novo Sequencing. The page returns peptide predictions with calculated masses and confidence scores, and you can save them as CSV or TSV.

For newcomers and experts: If you are new, click the built-in sample, run it and read the table that comes back. It lists peptide predictions, calculated masses and confidence scores. Try your own file once the sample makes sense.

If you know the field, decoding runs in parallel across the discrete amino acid states, and exact dynamic programming supplies reachability constraints. The Space offers three checkpoints. One is a production model labelled HC-PT plus nine species, one is pretrained on MassIVE-KB human data at 28M, and one has an extended vocabulary for post-translational modifications.

You upload a spectrum file or pick the sample
You choose a model checkpoint
The model decodes the peptides in parallel
You read the results and download CSV or TSV

Good to know: The Space is a model demo, so treat the predictions as research output.

• Models: facebook/esm2_t6_8M_UR50D and joelinator/dflow-novo-model
• Datasets: chrisagrams/massive_kb_v1_shuffled and InstaDeepAI/ms_proteometools
• Origin: the page cites a 2026 master's thesis from AIMS South Africa
• Version: Gradio 5.50.0
• Size: 43 files in the repository
• License: apache-2
• Cost to run: none named
• Data it reaches: the spectrum files you upload and the models it names, none else named
• Not for: no forbidden uses are named

On Hugging Face, 13 November 2026

• Likes: 1
• Built with: gradio
• License: apache-2.0

Links

• Space: https://huggingface.co/spaces/joelinator/dflow-novo
• Files: https://huggingface.co/spaces/joelinator/dflow-novo/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.