For each motif, calculate it's similarity to all k-mers of
length kmerLen, defined as the maximal probability of observing the
k-mer given the base frequencies of the motif (the maximum is taken over
for all possible ungapped alignments between motif and k-mer). If necessary
matrices are padded on the sides with background base frequencies (assuming
all bases to have a frequency of 0.25).
motifKmerSimilarity(
x,
kmerLen = 5,
kmers = NULL,
includeRevComp = FALSE,
BPPARAM = SerialParam(),
verbose = FALSE
)Either a PFMatrixList, or a character
scalar with a file containing motifs in HOMER format (used directly
method = "HOMER", loaded into a
PFMatrixList by homerToPFMatrixList
for method = "R").
A numeric scalar giving the k-mer length.
Either a character vector of k-mers for which to calculate
the similarity to each motif, or NULL, in which case all k-mers
of length kmerLen are used.
A logical scalar. If set to TRUE, each
k-mer as well as its reverse complement is compared to each motif, and the
larger of the two similarities is returned.
An optional BiocParallelParam
instance determining the parallel back-end to be used during evaluation.
A logical scalar. If TRUE, report on progress.
A matrix of probabilties for each motif - k-mer pair.
bplapply used for parallelization.
m <- rbind(A = c(12, 0, 0),
C = c( 3, 2, 0),
G = c( 0, 14, 0),
T = c( 0, 0, 15))
pfms <- TFBSTools::PFMatrixList(
TFBSTools::PFMatrix(name = "m1", profileMatrix = m),
TFBSTools::PFMatrix(name = "m2", profileMatrix = m[, 3:1])
)
motifKmerSimilarity(pfms, kmerLen = 3)[, c("AGT", "TGA")]
#> AGT TGA
#> m1 0.7000 0.0625
#> m2 0.0625 0.7000