genome-sequence-matching-for-nucleotide-patterns

Genome-sequence matching for nucleotide patterns

Summary

Match short nucleotide sequences (kmers or transcription factor motifs) to a reference genome to generate binary annotation matrices for downstream chromatin accessibility analysis. This skill converts genomic annotations into features suitable for deviation and variability computation in sparse ATAC-seq data.

When to use

You have filtered peak or chromatin accessibility counts and need to annotate each peak with the presence or absence of specific DNA sequence patterns—either predefined motifs (e.g., JASPAR transcription factor motifs) or kmers of a chosen length (6, 7, or higher)—to test whether particular sequences are associated with chromatin accessibility variability across cells or samples.

When NOT to use

• Peak coordinates are not already defined or filtered; use filterPeaks and peak-calling workflows first.
• You are performing de novo motif discovery rather than matching known motifs/kmers to peaks.
• Your input is RNA-seq or gene expression data rather than ATAC/DNAse-seq counts; chromatin accessibility annotation is not applicable.
• You already have precomputed deviation scores or a finalized feature table and do not need to re-annotate peaks.

Inputs

• Filtered chromatin accessibility counts object (SummarizedExperiment with peaks × samples/cells matrix)
• Reference genome (BSgenome object, e.g., BSgenome.Hsapiens.UCSC.hg19)
• Motif collection (PWM list or GRanges, e.g., from getJasparMotifs) OR kmer length integer (6, 7, etc.)

Outputs

• Binary annotation matrix (features × peaks): rows are kmers or motifs, columns are peaks, values are 0/1
• Motif index or kmer index object (passed to computeDeviations for deviation scoring)

How to apply

Load a reference genome (e.g., BSgenome.Hsapiens.UCSC.hg19) and your filtered peak/counts object. For motif matching, retrieve motifs from a database (e.g., JASPAR using getJasparMotifs) and apply matchMotifs from the motifmatchr package to identify which peaks contain which motifs, producing a binary matrix. Alternatively, for kmer-based annotation, call matchKmers with your desired kmer length (e.g., 6 or 7) on the counts object and genome, which returns a kmer annotation index. The resulting matrix rows correspond to kmers/motifs and columns correspond to peaks; entries are 1 if the pattern is present, 0 otherwise. This annotation matrix is then passed to computeDeviations to compute per-sample/cell deviation scores relative to the overall kmer/motif frequency.