Comparisons can be defined during the experiment design process and after the experiment had finished processing.
When setting up comparisons on ROSALIND you can drag and drop individual samples or sample attributes into each group. When using sample attributes to select attributes, ROSALIND looks for samples that satisfy all of the attribute requirements in each group. The number on the bottom of each group box shows the number of samples that qualify.
An optional covariate selection is available in Gene Expression experiments to correct for confounding variables. ROSALIND intelligently determines which experiment attributes may be valid covariates, reducing complexity.