Determining significance in metagenomic samples

Many ecology studies employ ordination methods to visually inspect metagenomic data sets, which initially may contain thousands of dimensions that represent operational taxonomic units (OTUs) of a sample. Many times, MANOVA (applied to a pairwise distance set) is applied to determine how different the groups in the study are from one another. It is convenient to have a p-value that allows us to interpret if two or more groups are different than one another with statistical confidence, where the null hypothesis is that the two populations are not different than the other. With MANOVA all groups are tested under the hypothesis that they are equal. In this work, we present a statistical framework for obtaining a p-value to compare multiple groups that is derived from a non-parametric statistical test, which uses data derived from the OTU features. The result is a matrix of p-values for the comparison on multiple groups in a metagenomic data set. We test our approach on a real-world database using several variations of ordination techniques.