calWeights: Calculate the weights for binding score

View source: R/calWeights.R

calWeightsR Documentation

Calculate the weights for binding score

Description

Use open score to calculate the weights for the binding score. The open score is calculated by the counts of the proximal region divided by the counts of the distal region. And the counts RangedSummarizedExperiment will be filtered by the Z-score of the open score. The weight is calculated by converting the Z score to the range of 0-1 following the normal distribution.

Usage

calWeights(se, openscoreZcutoff = 0, ...)

Arguments

se

An RangedSummarizedExperiment object. Outputs of countsNormalization.

openscoreZcutoff

Open score Z value cutoff value. Default is 0. Open score is calculated by the count ratio of proximal site and distal site.

...

Not used.

Value

A RangedSummarizedExperiment object with assays of count matrix with bindingSites, proximalRegion and distalRegion as column names and bindingSites GRanges object with the weights as rowRanges.

Author(s)

Jianhong Ou

Examples

bam <- system.file("extdata",
                   "KD.shift.rep1.bam",
                   package="ATACseqTFEA")
bsl <- system.file("extdata", "bindingSites.rds",
                   package="ATACseqTFEA")
bindingSites <- readRDS(bsl)
## get the count regions
bsEx <- expandBindingSites(bindingSites)
## count reads by 5'ends
res <- count5ends(bam, positive=0L, negative=0L,
                  bindingSites=bindingSites,
                  bindingSitesWithGap=bsEx$bindingSitesWithGap,
                  bindingSitesWithProximal=bsEx$bindingSitesWithProximal,
                  bindingSitesWithProximalAndGap=
                      bsEx$bindingSitesWithProximalAndGap,
                  bindingSitesWithDistal=bsEx$bindingSitesWithDistal)
## filter 0 counts in proximal
se <- eventsFilter(res, proximalRegion>0)
## normalize counts by width of count region
se <- countsNormalization(se, proximal=40, distal=40)
## calculate the weights
calWeights(se)

jianhong/ATACseqTFEA documentation built on Nov. 5, 2024, 9:46 a.m.