ttest

Synopsis

Perform various two-sided statistical analyses of data that is divided into two groups, including the Student's t-test. Each row (gene) in the data file is given a p value.

 ttest
	[-r: format line needs to be removed
   -w: use Welch approximate t
   -m; do mann-whitney U (a.k.a. Wilcoxon) test instead (this is experimental)
   -rank: use rank transformation of the data for ttest
   -l: log transform the data]  
	<data> <layout>

ttest -r -rank affydatafile.txt affydatafile-layout | /usr/local/bin/sort -gk 3 >! test.rank.out

(The -rank option typically yields many pvalues that are equal)

Inputs

• data: A tab-delimited data file, where each row represents a set of measurements to be analyzed. A p value is generated for each row in the file. See the detailed instructions for the format.
• layout: A simple file describing the experimental design. See the documentation of the format.

Outputs

The following columns

• The gene identifier
• The statistic ('t', or 'u' if the M-W test was used)
• The two-sided p value.
• The 'fold change' between the two groups. This is provided as a convenience and is not directly used in the analysis.

Options

• -r: The data file includes an extra line after the first line. (See the data format page for an explanation)
• -rank: Use the rank-transformation of the data. The ranks are used instead of the raw data. (a nonparametric version of the t-test)
• -m: EXPERIMENTAL: Do the Mann-Whitney (Wilcoxin) test instead (a non-parametric test). Note that in the current implementation the pvalues this yields are not very accurate for small numbers of samples.
• -l: Use the log transformation of the data. Do not use this if your data includes non-positive values.
• -w: Use the Welch 'approximate t' (applied when the variance in the two groups are not equal).

Dependencies

• Stats.pm
• This isn't a dependency, but gnu sort is useful for processing the output.

Problems/bugs

• The U test is approximate and is not very accurate for small numbers of samples (less than 20 or so)

Version history

References