calcTStatFast {sigPathway} | R Documentation |
Computes t-statistics and corresponding p-values.
calcTStatFast(tab, phenotype, ngroups = 2)
tab |
a numeric matrix of expression values, with the rows and columns representing probe sets and sample arrays, respectively |
phenotype |
a numeric (or character if ngroups >= 2)
vector indicating the phenotype |
ngroups |
an integer indicating the number of groups in the expression matrix |
If there are two groups in the matrix, it is recommended to use 0 and 1 to denote which sample columns belong to which group. If the phenotype is a character vector, then the phenotype ranked first in the alphabet is considered as 0.
If ngroups
= 2, the t-test done here is equivalent to a unpaired
two-sample t-test, assuming unequal variances. Please note that as of
version 1.1.6, the sign of the t-statistic is positive when the mean
of group 1 is greater than the mean of group 0.
If there is only one group in the matrix (e.g., Alzheimer's data set as reanalyzed in Tian et al. (2005)), then the phenotype vector should consist of continuous values. In this case, the association between phenotype and expression values is first calculated as Pearson correlation coefficients, transformed to Fisher's z, and then rescaled so that its variance is 1:
z = 0.5*log((1+rho)/(1-rho))*sqrt(n-3), where n is the number of phenotypes.
If ngroups
> 2, the f-statistics (from 1-way ANOVA) are
calculated. The user will need to check that the data have similar
variances among the groups.
pval |
A vector of unadjusted p-values |
tstat |
A vector of t-statistics (ngroups = 2) or rescaled
Fisher's z (ngroups = 1) |
rho |
(Also returned when ngroups = 1) A vector of Pearson
correlation coefficients |
Weil Lai
## Load inflammatory myopathy data set data(MuscleExample) statList <- calcTStatFast(tab, phenotype, ngroups = 2) ## Generate histogram of p-values hist(statList$pval, xlab = "Unadjusted p-values", ylab = "Frequency")