imageThis figure, which originally appeared in a poster presented at the American Society of Human Genetics, summarizes a MAMA analysis. The rows that are marked “combined,” “mean” and “covariance” and that are filled with the color red indicate strong relationships between phenotype and the genomic marker (p 0.≤001).

MAMA is a multivariate likelihood ration test that simultaneously assesses mean and covariance matrix difference among a set of variables. This methodology, developed by Nicholas J. Schork, Ph.D., Director of Biostatistics and Bioinformatics at STSI, 
 considers the fact that groups (e.g., subjects broken down into haplotype or genotype categories) can differ with respect to either mean values of a set of traits or the relationships between those variables.

The proposed tests, therefore, are more comprehensive and useful than are the many traditional multivariate tests that do not leverage information about both means and covariance matrices in their construction. The proposed tests are modifications and extensions of multvariate ANOVA and other techniques. A simultaneous test of the equality of means and covariance matrices across the genotypic categories can be constructed as simply the product of the test of means and covariances.