bolinas-dna/zoonomia-v1-v3_ccre_non_promoter

该数据集是跨哺乳动物训练集bolinas-dna/zoonomia-v1-v1的一个子集，专门针对标记为ccre_non_promoter（ENCODE cCRE V4非启动子区域）的锚点区域。这些区域包括dELS、pELS、CA等非启动子类别，但不包括启动子样信号（PLS），因为PLS区域已包含在其他子集中。数据集是六个v3子集分区的一部分，通过优先级顺序划分（cds > utr3 > ncrna_exon > tss_region_and_utr5 > ccre_non_promoter > background），每个锚点只分配一个标签。本子集包含468,131个人类锚点（占v1的41.18%），通过halLiftover投影到108种Zoonomia哺乳动物并进行反向互补增强后，扩展到96,639,800个训练样本。数据格式包括查询名称、物种、染色体位置、255 bp的序列（链感知）和增强信息等列。数据集用于基因组学研究和机器学习训练。This dataset is a subset of the cross-mammal training set bolinas-dna/zoonomia-v1-v1, specifically restricted to anchor regions labeled as ccre_non_promoter (ENCODE cCRE V4 non-promoter regions). These regions include non-promoter classes such as dELS, pELS, CA, etc., but exclude promoter-like signals (PLS) as PLS regions are already captured in other subsets. It is part of six v3 subset partitions, divided by priority order (cds > utr3 > ncrna_exon > tss_region_and_utr5 > ccre_non_promoter > background), with each anchor assigned exactly one label. The subset contains 468,131 human anchors (41.18% of v1), expanding to 96,639,800 training samples after halLiftover projection to up to 108 Zoonomia mammals and reverse-complement augmentation. The schema includes columns such as query name, species, chromosome positions, a 255 bp strand-aware sequence, and augmentation information. The dataset is designed for genomics research and machine learning training.