http://qiime.org/scripts/group_significance.html

This script is used to compare OTU frequencies in sample groups and to ascertain whether or not there are statistically significant differences between the OTU abundance in the different sample groups.

Output only makes sense for groups, not continuous variables !

The script will compare each OTU based on the passed sample groupings to see if it is differentially represented. The sample groupings are determined by the -c option. The script will group together samples which have the same value in the mapping file under the header passed with the -c option. Any samples that do not contain a value under the given header will not be included in the comparison. At a basic level, the script is constructing a OTUxSample (rowXcolumn) contingency table, and testing whether or not each OTU is differentially represented in cerstain groups of columns (determined by the metadata category passed).

We perform a Kruskal-Wallis test. Also availabe would be nonparametric_t_test, bootstrap_mann_whitney_u, ANOVA, g_test, parametric_t_test, mann_whitney_u.

Kruskal-Wallis: The null hypothesis is that the location paramater of the * groups of abundances for a given OTU is the same. The alternate hypothesis is * that at least one of the location parameters is different. *

kruskal_wallis - nonparametric ANOVA. This test is functionally an expansion of ANOVA to cases where the sample means are unequal and the distribution is not normal. The assumption that the distribution from which each group (within a single OTU) came is the same remains. This is a nonparametric test.

Output:

This script generates a tab separated output file with the following headers. OTU - OTU id Test-Statistic - the value of the test statistic for the given test P - the raw P value returned by the given test. 
FDR_P - the P value corrected by the Benjamini-Hochberg FDR procedure for multiple comparisons.
Bonferroni_P - the P value corrected by the Bonferroni procedure for multiple comparisons.
groupX_mean - there will be as many of these headers as there are unique values in the mapping file under the category passed with the -c option. Each of these fields will contain the mean frequency/abundance/count of the given OTU for the given sample group.
Taxonomy - this column will be present only if the biom table contained Taxonomy information. It will contain the taxonomy of the given OTU. 
