Taykhoom/

chromatin-activity-dart-eval

This repository contains the hg38 parquet release of Task 4 from DART-Eval. The task is to predict quantitative DNase accessibility from a 2,114 bp DNA sequence in one of five cell lines: GM12878, H1ESC, HEPG2, IMR90, or K562. The target is log1p of the summed unstranded DNase cut count over the central 1,000 bp of each sequence. Rows cover called peaks, GC-matched nonpeaks, and a separate set of IDR peaks. Published classification metrics… See the full description on the dataset page: https://huggingface.co/datasets/Taykhoom/chromatin-activity-dart-eval.