csaw (version 1.6.1)

consolidateClusters: Consolidate DB clusters

Description

Consolidate DB results from multiple analyses with cluster-level FDR control.

Usage

consolidateClusters(data.list, result.list, equiweight=TRUE, ...)

Arguments

data.list

a list of RangedSummarizedExperiment and/or GRanges objects

result.list

a list of data frames containing the DB test results for each entry of data.list

equiweight

a logical scalar indicating whether equal weighting from each analysis should be enforced

...

arguments to be passed to clusterWindows

Value

A named list is returned containing:

id

a list of integer vectors indicating the cluster ID for each window in data.list

region

a GRanges object containing the coordinates for each cluster

FDR

a numeric scalar containing the cluster-level FDR estimate

Details

This function consolidates DB results from multiple analyses, typically involving different window sizes. The aim is to provide comprehensive detection of DB at a range of spatial resolutions. Significant windows from each analysis are identified and used for clustering with clusterWindows. This represents the post-hoc counterpart to consolidateSizes.

Some effort is required to equalize the contribution of the results from each analysis. This is done by setting equiweight=TRUE, where the weight of each window is inversely proportional to the number of windows from that analysis. These weights are used as frequency weights for window-level FDR control (to identify DB windows prior to clustering). Otherwise, the final results would be dominated by large number of small windows.

Users can cluster by the sign of log-fold changes, to obtain clusters of DB windows of the same sign. However, note that nested windows with opposite signs may give unintuitive results - see mergeWindows for details.

See Also

clusterWindows, consolidateSizes

Examples

# Making up some GRanges.
low <- GRanges("chrA", IRanges(runif(100, 1, 1000), width=5))
med <- GRanges("chrA", IRanges(runif(40, 1, 1000), width=10))
high <- GRanges("chrA", IRanges(runif(10, 1, 1000), width=20))

# Making up some DB results.
dbl <- data.frame(logFC=rnorm(length(low)), PValue=rbeta(length(low), 1, 20))
dbm <- data.frame(logFC=rnorm(length(med)), PValue=rbeta(length(med), 1, 20))
dbh <- data.frame(logFC=rnorm(length(high)), PValue=rbeta(length(high), 1, 20))
result.list <- list(dbl, dbm, dbh)

# Consolidating.
cons <- consolidateClusters(list(low, med, high), result.list, tol=20)
cons$region
cons$id
cons$FDR

# Without weights.
cons <- consolidateClusters(list(low, med, high), result.list, tol=20, equiweight=FALSE)
cons$FDR

# Using the signs.
cons <- consolidateClusters(list(low, med, high), result.list, tol=20, fc.col="logFC")