kmer-motif-synergy-assessment

kmer-motif-synergy-assessment

Summary

Quantify redundancy and synergy between kmer and motif annotation sets in chromatin accessibility data using correlation and variance-based synergy metrics. This skill reveals which annotation types capture overlapping or complementary variability signals in chromVAR deviation objects.

When to use

After computing deviations for both motif and kmer annotations on the same chromVAR dataset, when you need to determine whether kmers and motifs are redundant predictors of chromatin accessibility variability or provide complementary information for downstream clustering, annotation, or feature selection.

When NOT to use

• Input deviations have not been bias-corrected and z-score normalized
• You only have a single annotation type (e.g., motifs only) and cannot compute pairwise relationships
• Annotation sets are already known to be identical or fully nested (correlation assessment would be redundant)

Inputs

• chromVARDeviations object with precomputed deviations and z-scores
• Motif annotation set (motif_ix columns)
• Kmer annotation set (kmer_ix columns)

Outputs

• Correlation matrix (CSV): Pearson correlations between motifs and kmers
• Synergy scores (CSV): annotation pair names, synergy z-scores, and p-values

How to apply

Load a chromVARDeviations object containing precomputed bias-corrected deviations and z-scores for both motif and kmer annotation sets. Subset the deviation matrix to isolate motif_ix and kmer_ix columns. Apply getAnnotationCorrelation to compute pairwise Pearson correlations between motifs and kmers across all samples, generating a correlation matrix with named rows and columns. Apply getAnnotationSynergy to compute z-scores for variability synergy by comparing the observed variance of peaks containing both annotation types against a random subsample of peaks with only the higher-variability annotation type. Export both matrices as named CSV files to preserve annotation identities and enable downstream interpretation.