Comparison of differential accessibility analysis strategies for ATAC-seq data

Abstract

ATAC-seq is widely used to measure chromatin accessibility and identify open chromatin regions (OCRs). OCRs usually indicate active regulatory elements in the genome and are directly associated with the gene regulatory network. The identification of differential accessibility regions (DARs) between different biological conditions is critical in determining the differential activity of regulatory elements. Differential analysis of ATAC-seq shares many similarities with differential expression analysis of RNAseq data. However, the distribution of ATAC-seq signal intensity is different from that of RNA-seq data, and higher sensitivity is required for DARs identification. Many different tools can be used to perform differential analysis of ATAC-seq data, but a comprehensive comparison and benchmarking of these methods is still lacking. Here, we used simulated datasets to systematically measure the sensitivity and specificity of six different methods. We further discussed the statistical and signal density cutoffs in the differential analysis of ATAC-seq by applying them to real data. Batch effects are very common in high-throughput sequencing experiments. We illustrated that batch-effect correction can dramatically improve sensitivity in the differential analysis of ATAC-seq data. Finally, we developed a user-friendly package, BeCorrect, to perform batch effect correction and visualization of corrected ATAC-seq signals in a genome browser. Gene regulation in the mammalian genome involves different types of regulatory elements, such as promoters, enhancers, and insulators. It was estimated that there are over two million regulatory elements in the human and mouse genomes 1,2 , and these regulatory elements recruit different epigenetic modifications to regulate the expression of genes in cell type-specific and developmental stage-specific manners 3-5. Active regulatory elements must remain in an accessible state to allow the binding of different transcription factors to activate or silence target genes. ATAC-seq (assay for transposase-accessible chromatin followed by sequencing) is a recently developed technique to measure genome-wide chromatin accessibility (or open chromatin) 6,7. Compared with other techniques, such as DNase-seq, Mnase-seq, and FAIRE-seq, ATAC-seq experiments are relatively easier to perform across different tissues and cell types. Furthermore, ATAC-seq experiments allow ultra-low input cell numbers, even down to the single-cell level 8. These advantages propelled ATAC-seq to be the most widely used technology to define open chromatin by many large genomics consortiums, including ENCODE 9 , TCGA 10 , PsychENCODE 11 , IHEC 12 , and TaRGET II 13. The peak-calling analysis used to identify open chromatin regions (OCRs) by using ATAC-seq is generally adapted from ChIP-seq data analysis. However, there are fundamental differences between ATAC-seq and ChIP-seq-most notably that ATAC-seq is performed without control or input samples. Nonetheless, peak callers, such as macs2 14 , can identify OCRs by evalua...

PeakVI: A Deep Generative Model for Single Cell Chromatin Accessibility Analysis

“…The bulk ATAC-seq data used as a ground truth reference for differential accessibility analysis was downloaded from GEO (accession GSE118189). We used the unstimulated samples of all B-cell and NK-cell subtypes included in the study and used DESeq2 [38], which was found to be among the ebst performing methods for differential accessibility from bulk ATAC-seq data [39] for differential accessibility between the two group. We then found regions in the hematopoiesis data that overlap with the regions in the bulk data, and used the differential signal found in the bulk data for the overlapping regions in the hematopoiesis data.…”

Tissue-specific usage of transposable element-derived promoters in mouse development

“…For each tissue, we calculated reads count table of peaks at E14.5 and P0 by using AIAP package and used edgeR to identify differential accessible regions (DARs) in the comparison between two development stages (abs (log2(Foldchange)) > 1 and FDR < 0.05, Additional file 5: FigureS10) [105,106]. Then, the accessible TEs overlapped with DARs were identified as differential accessible TEs between two stages in five tissues: (1) those accessible TEs with positive foldchange were defined as P0day-specific TEs that were more open at P0; and (2) accessible TEs with negative foldchange were defined as E14.5day-specific TEs that were more open at E14.5.…”

Bibliometric Analysis of ATAC-Seq and Its Use in Cancer Biology via Nucleic Acid Detection

“…For systematic ATAC-seq data analysis, Wei et al developed esATAC, which covers the elementary steps for full a full analysis procedure (9). Gontarz et al compared the sensitivity and specificity of six different accessibility analysis strategies for ATAC-seq data (10). In addition to chromosomal DNA, scientists can now use ATAC-seq to identify thousands of extrachromosomal circular DNA present in normal and tumor cells (11).…”