kmer-annotation-matrix-assembly

kmer-annotation-matrix-assembly

Summary

Generate k-mer annotation matrices that map short DNA sequence motifs (6-mers or 7-mers) to peaks in chromatin accessibility data, enabling subsequent deviation and variability scoring in chromVAR workflows. This is a prerequisite for motif-agnostic annotation of chromatin accessibility variability.

When to use

You have filtered peak counts from ATAC or DNase-seq data (with GC bias correction and sample/peak filtering applied) and want to annotate peaks by k-mer content rather than known transcription factor motifs—particularly when comparing how k-mer size affects the magnitude of chromatin accessibility variability scores.

When NOT to use

• You already have known transcription factor motif annotations and want to use matched motifs instead of k-mers.
• Your peak set contains overlapping peaks—apply filterPeaks() first to remove overlaps before k-mer annotation.
• You lack a reference genome object appropriate for your organism and assembly.

Inputs

• SummarizedExperiment counts object (filtered, GC-bias corrected, with min_depth ≥ 1500 and min_in_peaks ≥ 0.15)
• Reference genome object (BSgenome, e.g., BSgenome.Hsapiens.UCSC.hg19)
• k-mer length integer (e.g., 6 or 7)

Outputs

• Sparse binary matrix of k-mer annotations (rows = k-mers, columns = peaks)
• chromVAR kmer_ix object suitable for input to computeDeviations()

How to apply

Load the filtered SummarizedExperiment counts object and call matchKmers() with the desired k-mer length (6, 7, or other integer) and a reference genome (e.g., BSgenome.Hsapiens.UCSC.hg19). The function returns a sparse binary matrix where rows are k-mers and columns are peaks, with 1 indicating presence of that k-mer in the peak sequence. Repeat for multiple k-mer sizes if performing comparative analysis (e.g., 6-mers vs. 7-mers) to assess how motif granularity affects downstream deviation computation and variability scoring. The resulting annotation matrix is then passed directly to computeDeviations() for deviation scoring.