PREDA_main: function performing the core of PREDA analysis

Description Usage Arguments Details Value Author(s) See Also Examples

Description

function performing the core of PREDA analysis

Usage

1
2
3
4
5
6
PREDA_main(inputDataForPREDA, outputGenomicAnnotationsForPREDA
=NULL, nperms = 10000, verbose = TRUE, parallelComputations =
FALSE, multTestCorrection = "fdr", permutePerChromosome = FALSE,
blocksize = 10, permuteStatisticSign = FALSE, smoothMethod =
"lokern_scaledBandwidth_repeated", force = FALSE,
lokern_scaledBandwidthFactor = 2, limit.analysis = NULL)

Arguments

inputDataForPREDA

A Data for PREDA object

outputGenomicAnnotationsForPREDA

A GenomicAnnotationsForPREDA object.

If NULL, GenomicsAnnotations for output data are obtained from inputDataForPREDA

nperms

Number of permutations performed in PREDA analysis.

verbose

Logical, if TRUE some messages are printed concenrning the advancement of the analysis.

parallelComputations

Logical, if TRUE Rmpi is used to spawn slave processes, thus using parallel computing to speedup the analysis.

multTestCorrection

Multiple testing correction that will be adopted to correct the statistic p-values. Possible values are "fdr", for benjamini and Hochberg multiple testing correction and "qvalue" for p-values correction performed with qvalue package.

permutePerChromosome

Logical, if TRUE data parmutations are perfored separatedly for each chromsoome. In most cases the default value (FALSE) is preferable to avoid biases related to specific chromosomes extreme alterations.

blocksize

A parameter used to tune parallel computations if parallelComputations is TRUE. This is actually the number of permutations performed on each slave process before every communication with master process.

This is useftul to reduce the numebr of network communications when slow communicatinos are established among slave processes.

permuteStatisticSign

Logical, if TRUE statistics signs are permuted instead of permuting data along chromsomal position.

smoothMethod

The deafault smoothing metod used in the PREDA_main function is lokern smoothing with scaled bandwidth, using a scaling factor equal to 2.

Possible values are "lokern", for standard lokern smoothing, "quantsmooth", "spline" and "runningmean.x", where x is a user defined value for the number of adjacent data points using for running mean smoothing.

force

Logical, if TRUE force skipping quantsmooth control on number of data points. Singe quantsmooth is very slow with a high number of inpuit data, a check stopping computation with more than 2000 data points in one or more chromosome was introduced. This aprameter allow skippin this security check.

lokern_scaledBandwidthFactor

Factor of scaling for lokern estimated bandwidths

limit.analysis

Vector (numeric or character representing analyses names) to limit the output of preda analysis to a subset of input analyses.

Details

See supplementary material about PREDA method

Value

If outputGenomicAnnotationsForPREDA is NULL, a PREDADataAndResults object is returned. Otherwise a PREDAResults object is returned instead

Author(s)

Francesco Ferrari

See Also

Supplementary information about PREDA method

Examples

1
#See examples in PREDA tutorial

PREDA documentation built on May 6, 2019, 2:07 a.m.