The output will be a new table that contains the as many rows as groups and as many columns sets as possible permutations between positive and negative controls times the number of groups.
The output is the multivariate Z' factor and the classification error, an additional measure of how good the positive and negative control measurments seperate. Additionally for each control there is the number of samples and the status. The computation of the covariance matrix requires as many observations as parameter (rows as columns). So if this criteria is not met, the algorithm tries to bootstrap the matrix. If there are less than 3 observations, no Zprime is calculated (untrustworthy).
The multivariate Z' factors are implemented acoording to the Paper from Anne Kümmel: "Integration of Multiple Readouts into the Z' Factor for Assay Quality Assessment" (http://jbx.sagepub.com/content/15/1/95)
Note that the nature of the distributions from positive and negative control should be similar. The best results are obtained if the distributions are multivariate normal.
- Type: Tablescreeninput screen to be QC'ed