Cell Ranger ATAC1.1, printed on 04/19/2024
The cellranger-atac count pipeline outputs a BED-like tabular file, where each line represents a unique ATAC-seq fragment captured by the assay. Each fragment is created by two separate transposition events, which create the two ends of the observed fragment. Each unique fragment may generate multiple duplicate reads. These duplicate reads are collapsed into a single fragment record.
The first three columns of the fragments file are defined as in the BED format, so the fragments file can be treated as BED file in many cases.
The BED interval of the fragment is obtained by adjusting the BAM alignment interval of the sequenced read-pair. The start of the interval is moved forward by 4bp from a left-most alignment position and backward 5bp from the right-most alignment position. The transposase cuts the two DNA strands with a 9bp overhang, and adjusted positions represent the center point between these cuts; this position is recorded as a cut site that represents a chromatin accessibility event.
The pipeline outs/ folder contains fragments.tsv.gz and fragments.tsv.gz.tbi. The fragments.tsv.gz contains one line per unique fragment, with tab-separated fields as described below. The data is block-gzipped to allow indexing and to save disk space. The fragments.tsv.gz.tbi file is a tabix index of the fragment intervals facilitating random access to records from an arbitrary genomic interval. The tabix index is created with --preset=bed.
Positions in the fragments.tsv.gz file, as in a BED file, are 0-based.
| Column Number | Name | Description |
|---|---|---|
| 1 | chrom | Reference genome chromosome of fragment |
| 2 | chromStart | Adjusted start position of fragment on chromosome. |
| 3 | chromEnd | Adjusted end position of fragment on chromosome. The end position is exclusive, so represents the position immediately following the fragment interval. |
| 4 | barcode | The 10x cell barcode of this fragment. This corresponds to the CB tag attached to the corresponding BAM file records for this fragment. |
| 5 | duplicateCount | The number of PCR duplicate read pairs observed for this fragment. Sequencer-created duplicates, such as Exclusion Amp duplicates created by the NovaSeq™ instrument are excluded from this count. |