R/intersect.R

In chipseq: chipseq: A package for analyzing chipseq data

## NOTE: not exported or used anywhere

## effective.glength <- function(IR1, IR2, width = 200L)
## {
    
##     ## Motivation: think of IR1 as ranges of singleton islands in a
##     ## sample.  IR2 represents similar ranges in a different sample.
##     ## Assuming random distribution, the chance that IR2[i] intersects
##     ## IR1 is, up to the assumption that elements of IR1 are well
##     ## separated, is p = sum(width(IR1) + width) / G, where width is
##     ## the width of IR2[i].  Since all our widths are the same, a
##     ## simple approximation is

##     ## p = 2 * width * length(IR1) / G

##     ## Then, the number of elements in IR2 that intersects, X ~ Bin(m, p)

##     ## where m = length(IR2).  This may be used to estimate G.

##     if (!is(IR1, "IRanges")) IR1 <- IRanges(start = as.integer(IR1 - width/2L), width = width)
##     if (!is(IR2, "IRanges")) IR2 <- IRanges(start = as.integer(IR2 - width/2L), width = width)
##     m <- IR2 %in% IR1
##     phat <- mean(m)
##     Ghat <- 2 * width * length(IR1) / phat
##     Ghat
## }


## effective.glength.byChr <-
##     function(IRL1, IRL2, chroms = intersect(names(IRL1), names(IRL2)))
## {
##     sapply(chroms,
##            function(chr) {
##                effective.glength(IRL1[[chr]],
##                                  IRL2[[chr]])
##            })
## }