R: Estimate False Discovery Rate within the relative...
estimateFDR
R Documentation
Estimate False Discovery Rate within the relative substitution frequency
support by integrating PAR-CLIP data and RNA-Seq data
Description
Estimate upper and lower bounds for the False Discovery Rate within the
relative substitution frequency (RSF) support by integrating PAR-CLIP data
and RNA-Seq data (current version makes use of unstranded RNA-Seq)
A GRanges object, corresponding to a count table as
returned by the getAllSub function
RNASeq
GRanges object containing aligned RNA-Seq reads as returned by
readSortedBam
substitution
A character indicating which substitution is induced by
the experimental procedure (e.g. 4-SU treatment - a standard in PAR-CLIP
experiments - induces T to C transitions and hence substitution = 'TC' in
this case.)
minCov
An integer defining the minimum coverage required at a genomic
position exhibiting a substitution. Genomic positions of coverage less than
minCov are discarded. Default is 20 (see Details).
span
A numeric indicating the width of RSF intervals to be considered
for FDR computation. Defauls is 0.1 (i.e. 10 intervals are considered
spanning the RSF support (0,1]
cores
An integer defining the number of cores to be used for parallel
processing, if available. Default is 1.
plot
Logical, if TRUE a dotchart with cluster annotations is produced
verbose
Logical, if TRUE processing steps are printed
...
Additional parameters to be passed to the plot function
Details
For details on the FDR computation, please see Comoglio, Sievers and Paro.
Value
A list with three slots, containing upper and lower FDR bounds, and
the total number of positive instances each RSF interval. If plot,
these three vectors are depicted as a line plot.
Note
The approach used to compute the upper bound for the FDR is very
conservative. See supplementary information in Comoglio et al. for details.
Author(s)
Federico Comoglio and Cem Sievers
See Also
readSortedBam, getAllSub
Comoglio F, Sievers C and Paro R (2015) Sensitive and highly resolved identification
of RNA-protein interaction sites in PAR-CLIP data, BMC Bioinformatics 16, 32.