mergeResultsList: Wrappers to combine multiple analyses

In LTLA/csaw: ChIP-Seq Analysis with Windows

Wrappers to combine multiple analyses

Description

Convenience wrappers around lower-level functions to easily obtain region-level results from multiple sets of windows, each with their own statistics.

Usage

mergeResultsList(ranges.list, tab.list=NULL, tol, equiweight=TRUE, 
    get.best=TRUE, merge.args=list(), combine.args=list(), 
    best.args=list())

overlapResultsList(ranges.list, tab.list=NULL, regions, equiweight=TRUE, 
    get.best=TRUE, overlap.args=list(), combine.args=list(), 
    best.args=list())

Arguments

ranges.list	A list of GRanges or RangedSummarizedExperiment objects, usually containing windows of varying sizes from separate calls to windowCounts.
tab.list	A list of data.frames of differential binding results, usually from separate analyses at differing window sizes. Each should contain one row per interval for the corresponding entry of ranges.list. If NULL, defaults to a list containing the mcols of each entry of ranges.list.
tol	A numeric scalar to be passed to mergeWindows.
equiweight	A logical scalar indicating whether windows should be weighted so that each analysis contributes equally to the combined p-values.
get.best	A logical scalar indicating whether getBestTest or getBestOverlaps should be run.
regions	A GRanges class containing reference regions to overlap with the windows.
merge.args	Further arguments to pass to mergeWindows.
overlap.args	Further arguments to pass to findOverlaps.
combine.args	Further arguments to pass to combineTests or combineOverlaps.
best.args	Further arguments to pass to getBestTest or getBestOverlaps.

Details

mergeResultsList calls mergeWindowsList to identify clusters of windows corresponding to regions. These clusters are used in combineTests and (optionally) getBestTest to obtain per-region statistics.

overlapResultsList calls findOverlapsList to identify all windows overlapping each reference region. These overlaps are used in combineOverlaps and (optionally) getBestOverlaps, where each set of overlapping windows for a reference region are used to define the latter's statistics.

Weighting is performed as described in mergeWindowsList and findOverlapsList. If equiweight=FALSE, weights are reported but are not used in the p-value calculations.

Value

Both functions will return a DataFrame containing regions, a GRanges containing intervals of clustered windows (for mergeResultsList) or the input regions (for overlapResultsList); combined, the DataFrame containing combined statistics for each region; and best, the DataFrame containing statistics for the best window in each region.

The metadata of the DataFrame will contain ranges, a GRanges object containing the concatenated intervals from all elements of ranges.list; tab, a data.frame containing the combined results from all tab.list; and weights, a numeric vector of weights per window (for mergeResultsList) or per overlap (for overlapsResultList).

For mergeResultsList, the metadata will also contain ids, an integer vector specifying the cluster ID in the output regions for each interval in ranges.

For overlapResultsList, the metadata will also contain overlaps, a Hits object specifying the overlaps between the input regions (query) and ranges.

Author(s)

Aaron Lun

See Also

mergeWindowsList, combineTests and getBestTest, for the functions that actually do the work in mergeResultsList.

findOverlapsList, combineOverlaps and getBestOverlaps, for the functions that actually do the work in overlapResultsList.

Examples

# Making up some data.
tab1 <- data.frame(logFC=rnorm(20), logCPM=rnorm(20), PValue=rbeta(20, 1, 2))
gr1 <- GRanges("chrA", IRanges(sample(1000, 20), width=40))

tab2 <- data.frame(logFC=rnorm(20), logCPM=rnorm(20), PValue=rbeta(20, 1, 2))
gr2 <- GRanges("chrA", IRanges(sample(1000, 20), width=40))

# Merging windows.
(merge.out <- mergeResultsList(list(gr1, gr2), list(tab1, tab2), tol=1))
merge.out$regions
merge.out$combined

# Overlapping windows with a reference.
ref <- GRanges("chrA", IRanges(c(10, 500, 900), width=20))
(olap.out <- overlapResultsList(list(gr1, gr2), list(tab1, tab2), regions=ref))
olap.out$regions
olap.out$combined

LTLA/csaw documentation built on Dec. 21, 2024, 1:10 a.m.