gc-content-bias-calculation

gc-content-bias-calculation

Summary

Compute and annotate GC content bias for peaks in chromatin accessibility data to enable bias-corrected deviation scoring. This preprocessing step accounts for systematic GC-dependent biases in ATAC-seq or DNAse-seq fragment counts before motif matching and deviation computation.

When to use

Apply this skill when you have a SummarizedExperiment object containing peak counts from single-cell or bulk ATAC-seq/DNAse-seq data and need to prepare it for unbiased motif deviation analysis. The skill is required before filterSamples, motif matching, and computeDeviations steps to ensure deviation scores reflect true biological variability rather than GC-driven sequencing artifacts.

When NOT to use

• Peak regions are not defined or rowRanges of the SummarizedExperiment is empty.
• Reference genome sequences are unavailable for your organism of interest.
• Input data is already bias-corrected by another method (e.g., pre-normalized counts).

Inputs

• SummarizedExperiment object with peak counts (rowRanges defined as GRanges, assays containing count matrix)
• BSgenome reference object (e.g., BSgenome.Hsapiens.UCSC.hg19)

Outputs

• SummarizedExperiment object with updated rowData containing 'bias' column (GC content fraction per peak)

How to apply

Load a reference genome (e.g., BSgenome.Hsapiens.UCSC.hg19) and pass it to the addGCBias() function along with your SummarizedExperiment object containing peak regions in rowRanges. The function computes the GC content fraction for each peak and adds a 'bias' column to rowData. This bias annotation is then used internally by subsequent functions (computeExpectations, getBackgroundPeaks, computeDeviations) to match peaks by GC content when generating background sets and computing expected accessibility, ensuring that deviation scores are normalized for GC-driven biases in chromatin accessibility.

Related tools

• chromVAR (Primary R package containing addGCBias() function and downstream functions that consume the bias annotation) — https://github.com/GreenleafLab/chromVAR
• BSgenome.Hsapiens.UCSC.hg19 (Reference genome package providing DNA sequences needed to compute GC content for each peak)
• SummarizedExperiment (Data container class that holds peak counts and rowData (including bias annotations))