The fragment file is a data format introduced by 10x Genomics for the storage of scATAC-seq data. Fragment files are defined as coordinate-sorted, block gzip-compressed (bgzip) and indexed browser-extensible data files with the following five columns: chromosome, start, end, cell barcode, PCR duplicate count. The start and end fields of the fragment file correspond to positions of the two Tn5 integration events that generated the sequenced DNA fragment. As the fragment file contains a deduplicated and near-complete representation of a single-cell chromatin experiment and existing tools are established to efficiently retrieve subsets of a fragment file that overlap a given set of genomic regions44, we utilized the fragment file format as the central disk-based data structure in the Signac framework and is the only requirement for running a single-cell data analysis using Signac.
