Name: Dunnett

Firstname: Charles

Title: Professor

Institution: McMaster University

Street: 1280 Main Street West

City: Hamilton

Zip-Code: L8S 4K1

Country: Canada

Phone: 905-525-9140

Fax: 905-522-0935


Authors: Charles Dunnett

Title: Drug Screening and the False Discovery Rate

Abstract: The purpose of drug screening is to classify compounds as active or inactive. Although run as a series of experiments, it is an ongoing program geared to screen very large numbers. The rates of false positives and false negatives are of primary interest. Thus PCER is more appropriate than FWER to control. See the bibliography by Federer [1] for references. With respect to the FDR, I draw attention to a paper by King [2] who proposed a procedure based on maximizing the expected ratio of true actives among the total number passed to the next stage of development. This is equivalent to minimizing the FDR. Also, it controls the PCER.

References: [1] Federer, W.T. Procedures and designs useful for screening material. Biometrics, 19 (1963): 553-589. [2] King, E.P. A statistical design for drug screening. Biometrics, 19 (1963): 429-440.