# Saluki Human mRNA Half-Life Benchmark (mRNABench): Transcript Sequences with Aggregated Half-Life from 39 Datasets

## Facts

- Provider: Agarwal and Kelley (Saluki); redistributed by the Morris lab (mRNABench)
- License: CC-BY-4.0 (https://zenodo.org/records/6326409)
- Size: 3 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/morrislab/rnahl-saluki-human
- Access: public, readable without an account
- DOI: https://doi.org/10.1186/s13059-022-02811-x
- Category: Genomics & Bioinformatics
- Source: https://huggingface.co/datasets/morrislab/rnahl-saluki-human
- Page: https://scigantic.com/archive/282bf123-e0dc-5abe-8df5-f183d84c4ff7
- Keywords: Genomics & Bioinformatics, benchmark, Parquet, Hugging Face

## About

Human mRNA half-life targets for sequence-based models, redistributed by the Morris lab as part of mRNABench from the Saluki dataset of Agarwal and Kelley (2022), Genome Biology 23:245, which aggregated 39 human and 27 mouse transcriptome-wide half-life experiments. This repository holds the human transcripts. The target is the first principal component of half-life measurements across the aggregated datasets; each transcript also carries a binary CDS track marking the first nucleotide of each codon and a binary splice track marking the 3' end of each exon, so models can use coding and exon structure along with the sequence. One Parquet file (about 20 MB), loadable with datasets or pandas. Supports training and benchmarking of RNA language models and mRNA stability predictors; a mouse companion repository exists. The Saluki data are licensed CC BY 4.0 on Zenodo (record 6326409); cite Agarwal and Kelley (2022) and mRNABench.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("morrislab/rnahl-saluki-human")
print(ds)
```

## Data structure

File types in sampled files: Parquet

Example file: `rnahl-saluki-human.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/282bf123-e0dc-5abe-8df5-f183d84c4ff7
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/282bf123-e0dc-5abe-8df5-f183d84c4ff7
- Over MCP: `get_schema_card` with id `282bf123-e0dc-5abe-8df5-f183d84c4ff7` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Genomics & Bioinformatics datasets](https://scigantic.com/topics/genomics-and-bioinformatics)
- [Benchmark datasets](https://scigantic.com/topics/benchmark)
