enumerateByFDR: filter a ciseStore instance using an FDR threshold

In gQTLstats: gQTLstats: computationally efficient analysis for eQTL and allied studies

Description

filter a ciseStore instance using an FDR threshold

Usage

enumerateByFDR(store, fdrsupp, threshold = 0.05, filter=force,
   ids=NULL, trimToUnit=TRUE)

Arguments

store	instance of ciseStore-class
fdrsupp	instance of FDRsupp-class
threshold	upper bound on FDR to be included
filter	The FDR can be computed for any association score. To return only records satisfying a given filter, supply the filter function here. It may be desirable to carry a filter function from the storeToFDR stage, and this may be considered in future versions.
ids	if NULL, process all results in store, otherwise limit attention to jobs with id values in ids
trimToUnit	plug-in FDR estimates can sometimes lie outside [0,1] owing to sparsity or defects of extrapolation; if this parameter is TRUE, estimated FDR values outside [0,1] are moved to the nearest boundary

Details

uses storeApply, which will use BiocParallel infrastructure when available

Value

A GRanges instance with store contents to which estFDR is appended for each range. The estFDR quantity is predicted using the GAM model held in the FDRsupp instance.

Examples

## Not run: 
require(geuvStore2)
require(gQTLBase)
st = makeGeuvStore2()
data(filtFDR)
filtEnum = enumerateByFDR( st, filtFDR,
  filter=function(x)x[which(x$mindist <= 500000 & x$MAF >= 0.05)] )
names(metadata(filtEnum))
filtEnum[order(filtEnum$chisq, decreasing=TRUE)[1:2]]

## End(Not run) # not really essential

gQTLstats documentation built on Nov. 8, 2020, 7:53 p.m.