Fisher discriminant analysis plot of mouse liver samples. Samples were scored according to the canonical variables determined by Fisher Discriminant Analysis (FD Analysis). Each canonical variable is defined as a weighted sum of 100 specific genes. To score a sample, the gene expression value is multiplied by an FD Analysis coefficient, called a loading, and the products from the 100 genes used in the analysis are summed to give the canonical variable score for the sample. F/WR: Fasting/Weight Reduced.