Seminar Notice<DIV>
<DIV></DIV><BR>
<DIV>Statistical Laboratory</DIV>
<DIV>Iowa State University</DIV><BR>
<DIV>DATE AND TIME:&nbsp; Monday, March 7, 2005, 4:10 p.m.</DIV><BR>
<DIV>PLACE: 319 Snedecor</DIV><BR>
<DIV>SPEAKER: Dan Nettleton, Department of Statistics, Iowa State 
University</DIV><BR>
<DIV>TITLE:&nbsp; Using Observed p-values to Estimate the Number of True Null 
Hypotheses When Conducting Many Tests</DIV><BR><BR>
<DIV>ABSTRACT</DIV><BR><BR>
<DIV>Mosig et al. (2001, Genetics 157, 1683-1698) proposed an intuitively 
appealing method for estimating the number of true null hypotheses in a multiple 
test situation.&nbsp; They presented an iterative algorithm that relies on the 
distribution of observed p-values to obtain their estimator.&nbsp; In this talk, 
I will characterize the limit of their iterative algorithm and describe how 
their estimator can be computed directly without iteration.&nbsp; I will compare 
the performance of the resulting simple estimator with other procedures for 
estimating the number of true null hypotheses from a collection of observed 
p-values.</DIV><BR>
<DIV>Estimation of the number of true null hypotheses plays a direct role in the 
estimation of false discovery rate (FDR).&nbsp; The FDR concept has been used in 
many modern applications involving hundreds or thousands of tests.&nbsp; I will 
show how the proposed estimator can be used to approximate FDR for some 
microarray experiments conducted at Iowa State University.</DIV><BR>
<DIV>COFFEE:&nbsp;&nbsp;&nbsp; 3:45 p.m., 104 Snedecor Hall</DIV><BR>
	<DIV>
		</DIV>
</DIV>