Taykhoom/footprinting-dart-eval

DART-Eval Task 2: Transcription Factor Motif Footprinting

This repository contains the parquet release of Task 2 from DART-Eval. The task tests whether a DNA model can recognize transcription factor motifs inserted into neutral sequence backgrounds.

The dataset uses 1,443 HOCOMOCO v12 motifs and 100 neutral 350 bp backgrounds. For each motif and background, DART-Eval provides a positive sequence with the motif consensus inserted at the center and a negative sequence with a shuffled version of that motif. Both forward and reverse-complement orientations are included. The neutral backgrounds are also retained.

This is a synthetic motif-insertion task. The neutral sequences originated from the hg38 cCRE processing workflow, but the emitted examples do not carry genomic coordinates.

Dataset size

Label	Rows
positive	288,600
negative	288,600
neutral	200
total	577,400

All rows belong to the DART-Eval test split.

Loading

from datasets import load_dataset

dataset = load_dataset("Taykhoom/footprinting-dart-eval", split="test")

Columns

Column	Description
split	Always test.
sequence	350 bp DNA sequence.
label	positive, negative, or neutral.
pair_id	Shared identifier for a matched positive and negative sequence. Neutral rows have their own identifiers.
background_index	Index of the neutral background sequence, from 0 through 99.
motif	HOCOMOCO motif identifier. Null for neutral rows.
reverse_complement	Whether the sequence is in reverse-complement orientation.

Processing

The parquet reproduces the final DART-Eval _v1 footprinting dataset and its canonical HDF5 file. The complete workflow is available at:

https://github.com/TaykhoomDalal/DART-Eval-Processing/tree/main/Footprinting

The original benchmark code is available at:

https://github.com/kundajelab/DART-Eval

Sources and citation

The transcription factor motifs come from HOCOMOCO v12:

Vorontsov, I. E. et al. HOCOMOCO in 2024: a rebuild of the curated collection of binding models for human and mouse transcription factors. Nucleic Acids Research (2023). https://doi.org/10.1093/nar/gkad1077

Please also cite DART-Eval:

https://arxiv.org/abs/2412.05430

@inproceedings{patel2024darteval,
  title = {DART-Eval: A Comprehensive DNA Language Model Evaluation Benchmark on Regulatory DNA},
  author = {Patel, Aman and Singhal, Arpita and Wang, Austin and Pampari, Anusri and Kasowski, Maya and Kundaje, Anshul},
  booktitle = {Advances in Neural Information Processing Systems},
  volume = {37},
  year = {2024},
  url = {https://proceedings.neurips.cc/paper_files/paper/2024/hash/71998bfc3217ffe1cca1ee084dfadadd-Abstract-Datasets_and_Benchmarks_Track.html}
}

License

This repository repackages data distributed with DART-Eval. It does not assign a new license to the underlying data. Users should follow the terms and attribution requirements of DART-Eval and HOCOMOCO.