High-quality ATAC-seq libraries contain sub-nucleosomal-sized fragments (<150 bp) from open chromatin regions. We recommend performing 2x50 bp paired-end sequencing for standard ATAC-seq libraries, though longer read lengths are also acceptable (2x100bp, 2x150bp). It is highly recommended to trim adapter sequences prior to alignment to ensure fragments shorter than the read length can map to the reference genome. See this article for guidelines on FASTQ trimming.
ENCODE recommends 50 million uniquely aligned reads for a standard ATAC-seq dataset with human samples. However, the actual sequencing depth requirement heavily depends on the sample complexity and experimental goals. The required read depth could vary between 20-200 million. To assess the specific read depth requirement, we recommend performing downsampling analysis and compare crucial metrics across multiple read depths (Figure 1).
For comparative analyses among multiple ATAC-seq datasets (e.g., across drug treatment conditions), it is also recommended to downsample each uniquely aligned BAM file to the level of the sample with the lowest read depth.
.jpg)
Figure 1. Downsampling analyses reveal sequencing metrics impacted by read depths. Data generated from a standard ATAC-seq reaction using 50k K562 nuclei.