6 ATAC QC

ATAC-seq quality control evaluates the chromatin accessibility library for each cell. The two key ATAC-specific metrics are nucleosome signal and TSS enrichment score. Nucleosome signal measures the ratio of mononucleosomal to nucleosome-free fragments — a low signal indicates a high proportion of accessible (open) chromatin reads, characteristic of a good quality library. TSS enrichment score measures the fold-enrichment of fragments at transcription start sites relative to flanking regions — high TSS enrichment confirms that the ATAC library is capturing accessible regulatory elements rather than background noise.

6.1 Compute ATAC QC metrics

Compute per-cell nucleosome signal and TSS enrichment score for each sample. These metrics are added to the Seurat object metadata alongside the RNA QC metrics computed in the previous chapter.

cntsB <- NucleosomeSignal(cntsB, assay = "ATAC")
cntsB <- TSSEnrichment(cntsB,    assay = "ATAC")

cntsC <- NucleosomeSignal(cntsC, assay = "ATAC")
cntsC <- TSSEnrichment(cntsC,    assay = "ATAC")

cntsD <- NucleosomeSignal(cntsD, assay = "ATAC")
cntsD <- TSSEnrichment(cntsD,    assay = "ATAC")

6.2 ATAC metrics per donor

Per-donor violin plots of nFeature_ATAC, nCount_ATAC, TSS.enrichment, and nucleosome_signal. Comparing donors within the same sample highlights donor-specific differences in ATAC library quality that could affect downstream peak calling or differential accessibility analysis.

6.2.1 Sample B

VlnPlot(cntsB, features = "nFeature_ATAC",     pt.size = 0.1, group.by = 'donor_id') + NoLegend()

VlnPlot(cntsB, features = "nCount_ATAC",       pt.size = 0.1, group.by = 'donor_id') + NoLegend()

VlnPlot(cntsB, features = "TSS.enrichment",    pt.size = 0.1, group.by = 'donor_id') + NoLegend()

VlnPlot(cntsB, features = "nucleosome_signal", pt.size = 0.1, group.by = 'donor_id') + NoLegend()

6.2.2 Sample C

VlnPlot(cntsC, features = "nFeature_ATAC",     pt.size = 0.1, group.by = 'donor_id') + NoLegend()

VlnPlot(cntsC, features = "nCount_ATAC",       pt.size = 0.1, group.by = 'donor_id') + NoLegend()

VlnPlot(cntsC, features = "TSS.enrichment",    pt.size = 0.1, group.by = 'donor_id') + NoLegend()

VlnPlot(cntsC, features = "nucleosome_signal", pt.size = 0.1, group.by = 'donor_id') + NoLegend()

6.2.3 Sample D

VlnPlot(cntsD, features = "nFeature_ATAC",     pt.size = 0.1, group.by = 'donor_id') + NoLegend()

VlnPlot(cntsD, features = "nCount_ATAC",       pt.size = 0.1, group.by = 'donor_id') + NoLegend()

VlnPlot(cntsD, features = "TSS.enrichment",    pt.size = 0.1, group.by = 'donor_id') + NoLegend()

VlnPlot(cntsD, features = "nucleosome_signal", pt.size = 0.1, group.by = 'donor_id') + NoLegend()

6.3 TSS enrichment vs ATAC feature count

Density scatter plot of nFeature_ATAC against TSS.enrichment. High-quality cells cluster in the upper-right region (many accessible peaks, high TSS enrichment). The quantile lines help identify the threshold below which cells fall outside the expected quality range.

6.3.1 Sample B

DensityScatter(cntsB, x = 'nFeature_ATAC', y = 'TSS.enrichment',
               log_x = TRUE, quantiles = TRUE)

6.3.2 Sample C

DensityScatter(cntsC, x = 'nFeature_ATAC', y = 'TSS.enrichment',
               log_x = TRUE, quantiles = TRUE)

6.3.3 Sample D

DensityScatter(cntsD, x = 'nFeature_ATAC', y = 'TSS.enrichment',
               log_x = TRUE, quantiles = TRUE)