Recent advances in FDR methodology
Daniel Yekutieli, Tel Aviv University

False Discovery Rate controlling procedures have proven to be essential tools in
dealing with large multiplicity problems. I will review existing FDR
methodology and introduce two new methodologies. FDR confidence intervals - a
new methodology for constructing valid confidence intervals for selected
parameters, and FDR trees ~V a hierarchical FDR controlling testing scheme in
which the number and identity of the hypotheses tested is determined by the
data itself. I will illustrate the use of the FDR methodologies by applying
them to genetic data: analysis of a complex microarray experiment and the
genetic mapping of multiple quantitative traits.

Joint work with Yoav Benjamini