Statistical methods for studying homozygous typing cells of unknown specificity.

A situation can arise in D typing in which the HLA (A,B,C,D, or DR) specificities of the responders are known and the specificities of the HTC's are unknown. The most powerful and direct method of detecting association between the unknown stimulator (HTC) and any given HLA specificity is by comparing the observed double normalized values (DNV's) of individuals known to be positive for the specificity with the DNV's of negative individuals. This can be done by comparing the two groups with a Kolmogorov-Smirnov test (K-S test), an established statistical procedure for evaluating correlation between continuous variables, such as the DNV, and discrete variables (such as presence of D type). The application of the K-S test will generate as a "cutoff" value of point that maximizes the average of the frequencies of correct assignments in D positives and D negatives. We also propose an alternative method of computing the "r" value. We have analyzed 49 HTC's from the 8th International Workshop and present the association observed with the D and DR specificities.