# PROSPECT PTMs: fragment ion intensity prediction (MS2) benchmark for modified peptides

## Facts

- Provider: Wilhelm Lab, Technical University of Munich
- License: CC-BY-4.0 (https://huggingface.co/datasets/Wilhelmlab/prospect-ptms-ms2)
- Size: 100 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/Wilhelmlab/prospect-ptms-ms2
- Access: public, readable without an account
- Category: Proteomics
- Source: https://github.com/wilhelm-lab/PROSPECT
- Page: https://scigantic.com/archive/7816cbbf-fb0d-5ff2-ae1c-abd3a995c5cf
- Keywords: Proteomics, benchmark, Parquet, Hugging Face

## About

A proteomics machine learning dataset from the Wilhelm lab (Technical University of Munich, School of Life Sciences), annotated, processed and split for fragment ion intensity (MS2 spectrum) prediction from peptide sequence, including post-translational modifications. It is derived from the PROSPECT PTMs datasets on Zenodo, which are built on the ProteomeTools synthetic peptide library measurements in PRIDE; fragment annotations follow the expert-system rules of the PROSPECT paper and intensities are stored in the intensities_raw column, with optional features for PTM encoding and acquisition setup such as collision energy and fragmentation. Two configurations: default (train, validation and test drawn from the PROSPECT PTMs sets) and holdout (the PROSPECT Test-PTM set, for final model comparison). Parquet under data/ and holdout/, about 4 GB; load with load_dataset("Wilhelmlab/prospect-ptms-ms2") or the "holdout" config. Suited to training Prosit-style spectrum predictors for PTM-aware peptide identification and rescoring. Curated by the Wilhelm lab under CC BY 4.0; cite the PROSPECT paper and the ProteomeTools papers.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("Wilhelmlab/prospect-ptms-ms2")
print(ds)
```

## Data structure

File types in sampled files: Parquet

Example file: `data/test-00000-of-00010.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/7816cbbf-fb0d-5ff2-ae1c-abd3a995c5cf
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/7816cbbf-fb0d-5ff2-ae1c-abd3a995c5cf
- Over MCP: `get_schema_card` with id `7816cbbf-fb0d-5ff2-ae1c-abd3a995c5cf` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Benchmark datasets](https://scigantic.com/topics/benchmark)
- [Proteomics datasets](https://scigantic.com/topics/proteomics)
