Skip to contents

Analyze pairwise and multi-way overlap between transcription factor binding sites

Source: R/tf_overlap.R
analyze_tf_overlap.Rd

For N TFs in the enhanceosome, computes pairwise overlap fractions, unique region counts per TF, shared region counts, and UpSet-style intersection data for all TF combinations. Uses Jaccard index for symmetric overlap quantification and overlap coefficient for asymmetric assessment (Church & Hanks, 1990). The presence/absence matrix pattern follows the approach used by DiffBind (Stark & Brown, 2011).

Usage

analyze_tf_overlap(enhanceosome, database, regions = NULL)

Arguments

enhanceosome

epiRomics class database containing enhanceosome calls, or an epiRomics database with ChIP data

database

epiRomics class database containing all data initially loaded

regions

GRanges object defining regions to analyze. If NULL, uses all enhanceosome annotation regions.

Value

list with components:

overlap_matrix

matrix of pairwise overlap coefficients (asymmetric: fraction of row TF overlapping col TF)

overlap_counts

matrix of pairwise absolute overlap counts

jaccard_matrix

matrix of pairwise Jaccard indices (symmetric: intersection/union)

unique_counts

named integer vector of regions bound by ONLY that TF

shared_all

integer count of regions bound by ALL TFs

n_tf_counts

table of how many regions are bound by exactly 1, 2, 3... N TFs

tf_names

character vector of TF names analyzed

summary

data.frame with per-TF summary statistics

References

  • Church KW, Hanks P (1990) Computational Linguistics 16(1):22-29. "Word Association Norms, Mutual Information, and Lexicography." Pointwise mutual information framework adapted for co-binding analysis.

  • Stark R, Brown GD (2011) DiffBind, Bioconductor. Presence/absence matrix approach for binding site overlap analysis.

Examples

db <- make_example_database()
eso <- make_example_enhanceosome(db)
overlap <- analyze_tf_overlap(eso, db)
overlap$overlap_matrix
#>     TF1    TF2
#> TF1   1 0.6667
#> TF2   1 1.0000