mincFDR: False Discovery Rates

Description Usage Arguments Details Value Methods (by class) See Also Examples

View source: R/minc_FDR.R

Description

Takes the output of a minc modelling function and computes False Discovery Rate thresholds.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
mincFDR(buffer, ...)

## S3 method for class 'mincSingleDim'
mincFDR(buffer, df, mask = NULL, method = "qvalue",
  ...)

## S3 method for class 'mincLogLikRatio'
mincFDR(buffer, mask = NULL, ...)

## S3 method for class 'mincLmer'
mincFDR(buffer, mask = NULL, method = "fdr", ...)

## S3 method for class 'mincMultiDim'
mincFDR(buffer, columns = NULL, mask = NULL,
  df = NULL, method = "FDR", statType = NULL, ...)

Arguments

buffer

The results of a mincLm type run.

...

extra parameters to pass to methods

df

The degrees of freedom - normally this can be determined from the input object.

mask

Either a filename or a numeric vector representing a mask only values inside the mask will be used to compute the threshold.

method

The method used to compute the false discovery rate. Options are "FDR" and "pFDR".

columns

A vector of column names. By default the threshold will be computed for all columns; with this argument the computation can be limited to a subset.

statType

This should be either a "t","F","u","chisq" or "tlmer" depending upon the type of statistic being thresholded.

Details

This function uses the qvalue package to compute the False Discovery Rate threshold for the results of a mincLm computation. The False Discovery Rate represents the percentage of results expected to be a false positive. Two implementations can be used as specified by the method argument. "FDR" uses the implementation in p.adjust, whereas "pFDR" is a version of the postivie False Discovery Rate as found in John Storey's qvalue package. The main interface functions are

Value

A object of type mincQvals with the same number of columns as the input (or the subset specified by the columns argument to mincFDR). Each column now contains the qvalues for each voxel. Areas outside the mask (if a mask was specified) will be represented by a value of 1. The result also has an attribute called "thresholds" which contains the 1, 5, 10, 15, and 20 percent false discovery rate thresholds.

Methods (by class)

See Also

mincWriteVolume,mincLm,mincWilcoxon or mincTtest

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
## Not run: 
getRMINCTestData() 
# read the text file describing the dataset
gf <- read.csv("/tmp/rminctestdata/test_data_set.csv")
# run a linear model relating the data in all voxels to Genotype
vs <- mincLm(jacobians_fixed_2 ~ Sex, gf)
# compute the False Discovery Rate
qvals <- mincFDR(vs)

## End(Not run)

Mouse-Imaging-Centre/RMINC documentation built on Oct. 5, 2018, 9:23 a.m.