winnow

KO | EN

gitlite — search

#python #hacktoberfest #ai #llm #machine-learning #rust #deep-learning #docker #macos #java #windows #awesome #cli #openai

(★ 10)

Confidence control and FDR estimation for de novo peptide sequencing

winnow Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<br />
<div align="center">
<h1 align="center">Winnow</h1>

<p align="center">
    Confidence calibration and FDR control for <i>de novo</i> peptide sequencing
    <br />
    <a href="https://instadeepai.github.io/winnow/"><strong>Explore the docs »</strong></a>
    <br />
    <br />
    <a href="https://github.com/instadeepai/winnow/issues/new?labels=bug&template=bug_report.md">Report bug</a>
    &middot;
    <a href="https://github.com/instadeepai/winnow/issues/new?labels=enhancement&template=feature_request.md">Request feature</a>
  </p>
</div>

<details>
  <summary>Table of Contents</summary>
  <ol>
    <li>
      <a href="#about-the-project">About the project</a>
    </li>
    <li>
      <a href="#installation">Installation</a>
    </li>
    <li>
      <a href="#usage">Usage</a>
    </li>
    <li><a href="#contributing">Contributing</a></li>
  </ol>
</details>

<div align="center">
  <img src="https://raw.githubusercontent.com/instadeepai/winnow/main/docs/assets/winnow_workflow.png" alt="Winnow Workflow" style="max-width:100%;">
  <p>Winnow workflow for confidence calibration and FDR control in <em>de novo</em> peptide sequencing</p>
</div>

## About the project

In bottom-up proteomics workflows, peptide sequencing—matching an MS2 spectrum to a peptide—is just the first step. The resulting peptide-spectrum matches (PSMs) often contain many incorrect identifications, which can negatively impact downstream tasks like protein assembly.

To mitigate this, intermediate steps are introduced to:

1. Assign confidence scores to PSMs that better correlate with correctness.
2. Estimate and control the false discovery rate (FDR) by filtering identifications based on confidence scores.

For database search-based peptide sequencing, PSM rescoring and target-decoy competition (TDC) are standard approaches, supported by an extensive ecosystem of tools. However, *de novo* peptide sequencing lacks standardised methods for these tasks.

Winnow aims to fill this gap by implementing the calibrate-estimate framework for FDR estimation. Unlike TDC, this approach is directly applicable to *de novo* sequencing models. Additionally, its calibration step naturally incorporates common confidence rescoring workflows as part of FDR estimation.

Winnow provides both a CLI and a Python package, offering flexibility in performing confidence calibration and FDR estimation.

## Installation

Winnow is published on PyPI as **`winnow-fdr`**. Install with `pip` or a `pip`-compatible tool (e.g. `uv pip install`):

```python
pip install winnow-fdr
```

```python
uv pip install winnow-fdr
```

## Quick Start

Get started in minutes using the example data in `examples/example_data/`.

```bash
# Train a calibrator on the example data
make train-sample

# Run prediction with the trained model without filtering on an FDR threshold
make predict-sample
```

**Note:** The sample data is minimal (100 spectra) and intended for testing only. The `make` commands shown above are configured for the sample data with adjusted settings (e.g., relaxed FDR threshold). For your own datasets, use the `winnow` commands outlined below.

## Usage

Winnow supports two usage modes:

1. A command-line interface (CLI) with sensible defaults and multiple FDR estimation methods.
2. A configurable and extensible Python package for advanced users.

### CLI

Installing Winnow provides the `winnow` command with four sub-commands:

1. `winnow train` – Performs confidence calibration on a dataset of annotated PSMs, outputting the fitted model checkpoint.
2. `winnow compute-features` – Computes and outputs the feature set for a dataset of PSMs.
3. `winnow predict` – Performs confidence calibration using a fitted model checkpoint (defaults to a pretrained general model from Hugging Face), estimates and controls FDR using the calibrated confidence scores.
4. `winnow diagnose-calibration` – On a labelled holdout, estimates tail expected calibration error (TECE) and signed tail expected calibration error (sTECE) at the FDR operating threshold and optionally writes a reliability diagram.

By default, `winnow predict` uses a pretrained general model ([`InstaDeepAI/winnow-general-model`](https://huggingface.co/InstaDeepAI/winnow-general-model)) hosted on Hugging Face Hub, allowing you to get started immediately without training. You can also specify custom Hugging Face models or use locally trained models.

Winnow uses [Hydra](https://hydra.cc/) for flexible, hierarchical configuration management. All parameters can be configured via YAML files or overridden on the command line:

```bash
# Quick start with defaults
winnow predict

# Override specific parameters
winnow predict fdr_control.fdr_threshold=0.01

# Specify different data source and dataset paths
winnow predict data_loader=mztab dataset.spectrum_path_or_directory=data/spectra.parquet dataset.predictions_path=data/preds.mztab
```

Refer to the [CLI Guide](https://instadeepai.github.io/winnow/cli/) and [Configuration Guide](https://instadeepai.github.io/winnow/configuration/) for details on usage and configuration options.

The [example notebook](https://github.com/instadeepai/winnow/blob/main/examples/getting_started_with_winnow.ipynb) walks through the Python API for the same workflows you can run from the CLI.

## Contributing

Contributions are what make the open-source community such an amazing place to learn, inspire and create, and we welcome your support! Any contributions you make are **greatly appreciated**.

If you have ideas for enhancements, you can:

- Fork the repository and submit a pull request.
- Open an issue and tag it with "enhancement".

### Contribution process

1. Fork the repository.
2. Create a feature branch (`git checkout -b feat-amazing-feature`).
3. Commit your changes (`git commit -m 'feat: add some amazing feature'`).
4. Push to your branch (`git push origin feat-amazing-feature`).

For more details on the contributing process, see the [Contributing Guide](https://instadeepai.github.io/winnow/contributing/).

Don't forget to give the project a star! Thanks again! :star:

## BibTeX entry and citation information

If you use Winnow in your research, please cite the following preprint:

```bibtex
@article{mabona2025novopeptidesequencingrescoring,
     title={De novo peptide sequencing rescoring and FDR estimation with Winnow},
     author={Amandla Mabona and Jemma Daniel and Henrik Servais Janssen Knudsen
     and Rachel Catzel and Kevin Michael Eloff and Erwin M. Schoof and Nicolas Lopez
     Carranza and Timothy P. Jenkins and Jeroen Van Goey and Konstantinos Kalogeropoulos},
      year={2025},
      eprint={2509.24952},
      archivePrefix={arXiv},
      primaryClass={q-bio.QM},
      url={https://arxiv.org/abs/2509.24952},
}
```

winnow

winnow Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

winnow

winnow Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README