R/getRandomEnrichment.R

# ---------------------------------------------------------------------------- #
#' get enrichment based on randomized feature overlap
#'
#' This function measures the association between two genomic features by randomizing one feature and counting the overlaps in randomized sets.
#' That is to say, \code{query} feature will be randomly distributed over the genome (constrained by provided options), and the overlap of \code{target} with these randomized features will be measured.
#'
#' @param target a \code{GRanges} object to be overlapped with \code{query}
#' @param query a \code{GRanges} object that will be randomly placed across the genome and overlap of these random regions with \code{target} will be the background distribution of association between \code{target} and \code{query}.
#' @param rand.set instead of randomly placing features in \code{query} one can supply an already shuffled set of \code{query} genomic features.
#' @param randomizations  number of times the features to be shuffled
#' @param ... other parameters to be passed to \code{randomizeFeature} function. These parameters ccontrol how randomization is done.
#' @return returns a \code{RandomEnrichment} object
#' @seealso \code{\link{randomizeFeature}}
#' @examples
#' data(cage)
#' data(cpgi)
#' \donttest{ 
#' enr = getRandomEnrichment(cage, cpgi, randomizations=50)
#' }
#' @export
#' @docType methods
#' @rdname getRandomEnrichment-methods
setGeneric("getRandomEnrichment", 
           function(target, query, randomizations=1000, rand.set=NULL,...) 
             standardGeneric("getRandomEnrichment") )

#' @aliases getRandomEnrichment,GRanges,GRanges-method
#' @rdname getRandomEnrichment-methods
setMethod("getRandomEnrichment",
            signature(target = "GRanges",query="GRanges"),
            function( target, query, randomizations, rand.set,... ){

                orig.cnt = sum(countOverlaps(target,query) > 0)

                if( is.null(rand.set) ){
                
                    rand.olap.dist=numeric(randomizations)
                    for(i in 1:randomizations){
#                         message("Iteration number:",i,"\r")
                        my.rand = randomizeFeature(query,...)
                        my.cnts = sum(countOverlaps(target,my.rand)>0)
                        rand.olap.dist[i] = my.cnts
                    }
#                     message("\n")
                    
                }else if(is(rand.set,"GRangesList")){
                    
                    randomizations = length(rand.set)
                    rand.olap.dist = numeric(randomizations)
                    for(i in 1:randomizations){
                        message("Iteration number:",i,"\r")
                        my.cnts = sum(countOverlaps(target,rand.set[[i]])>0)
                        rand.olap.dist[i] = my.cnts
                    }
                    message("\n")
                }else{
                    stop("Wrong 'rand.set' argument supplied, it should be a 'GRangesList'")
                }

                # calculate p-value assuming normal distribution
                p.value=pnorm(orig.cnt, 
                      mean=mean(rand.olap.dist), 
                      sd=sd(rand.olap.dist),
                      lower.tail=FALSE )
                #if(p1>0.5){p.value=1-p1}else{p.value=p1}

                # calculate randomized raw p-value
                rand.p.value = sum(rand.olap.dist>orig.cnt)/length(rand.olap.dist)

                n = new('RandomEnrichment', 
                orig.cnt = orig.cnt,
                          rand.olap.dist = rand.olap.dist,
                        log2fc = log2(orig.cnt/mean(rand.olap.dist)),
                          p.value = p.value,
                          rand.p.value = rand.p.value)
        return(n)
})

Try the genomation package in your browser

Any scripts or data that you put into this service are public.

genomation documentation built on Nov. 8, 2020, 5:21 p.m.