PEAKQC: Periodicity Evaluation in scATAC-seq data for quality assessment

Abstract

ATAC-seq is a common protocol to identify regulatory regions in the genome. While quality control (QC) standards are well-established for bulk ATAC-seq, application to single-cell ATAC-seq remains challenging due to data sparsity, noise, and a lack of consensus on effective QC metrics and thresholds. Existing methods, such as fragment length ratio offer limited resolution and fail to fully utilize fragment length distribution (FLD) patterns. We present PEAKQC, a Python-based tool designed to assess single-cell ATAC-seq data quality using a wavelet-based convolution of FLD patterns. Benchmarking PEAKQC against established QC metrics demonstrates an overall more consistent, linear quality assessment of ATAC data. PEAKQC improves downstream analyses by effectively filtering low-quality cells while preserving biologically meaningful data. When combined with other metrics, such as the ratio of fragments in peaks and total counts, PEAKQC enhances clustering accuracy and cell-type identification. The tool is modular, easily installable via Python Package Index, and integrates seamlessly into Python-based single-cell analysis frameworks. PEAKQC provides a robust and scalable solution for single-cell ATAC-seq QC, addressing current gaps in the field and suggesting FLD patterns as a new standard for data quality assessment.