With over 200 drugs and 56 cancer sites studied, we generated over 11,000 comparisons with each over-all screening analysis.  Whether or not we modify our statistical significance testing to account for multiple comparisons, such as by Bonferroni correction, is controversial.  We are certainly concerned with the many chance associations that occur, but we do not want to miss important relationships.  What we have done is, in many situations, lower the p value of interest from 0.05 to 0.01.  But more important, we emphasize in our reports that many of the associations that we observe are merely due to chance.