waldTest: Perform gene-wise Wald test for two group comparisons for...

In haowulab/DSS: Dispersion shrinkage for sequencing data

Perform gene-wise Wald test for two group comparisons for sequencing count data.

Description

The counts from two groups are modeled as negative binomial random variables with means and dispersions estimated. Wald statistics will be constructed. P-values will be obtained based on Gaussian assumption.

Usage

## S4 method for signature 'SeqCountSet'
waldTest(seqData, sampleA, sampleB, equal.var, fdr.method=c("BH", "locfdr"))

Arguments

seqData	An object of SeqCountSet class.
sampleA	The sample labels for the first sample to be compared in two-group comparison.
sampleB	The sample labels for the second sample to be compared in two-group comparison.
equal.var	A boolean to indicate whether to use the same or different means in two groups for computing variances in Wald test. Default is FALSE.
fdr.method	Method to compute FDR. Availabile options are "BH" for Benjamini-Hochberg FDR method, or local FDR from "locfdr" package.

Details

The input seqCountData object Must have normalizationFactor and dispersion fields filled, e.g., estNormFactors and estDispersion need to be called prior to this. With group means and shrunk dispersions ready, the variances for difference in group means will be constructed based on Negative Binomial distribution. P-values will be obtained under the assumption that the Wald test statistics are normally distributed. Genes with 0 counts in both groups will be assigned 0 for test statistics and 1 for p-values.

Value

A data frame with each row corresponding to a gene. Rows are sorted according to wald test statistics. The columns are:

gene Index	index for input gene orders, integers from 1 to the number of genes.
muA	sample mean (after normalization) for sample A.
muB	sample mean (after normalization) for sample B.
lfc	log fold change of expressions between two groups.
difExpr	differences in expressions between two groups.
stats	Wald test statistics.
pval	p-values.
fdr	FDR.
local.fdr	Local FDR if the FDR method is "locfdr".
others	input gene annotations supplied as AnnotatedDataFrame when constructed the SeqCountData object.

Author(s)

Hao Wu <[e-mail]>

Examples

data(seqData)
seqData=estNormFactors(seqData)
seqData=estDispersion(seqData)
result=waldTest(seqData, 0, 1)
head(result)

haowulab/DSS documentation built on Oct. 28, 2023, 6:59 p.m.