False Discovery Rate and Related Analyses for Scan Statistics
David O. Siegmund
Statistics, Stanford University, Stanford, CA, United States

Motivated by problems of signal detection in biology, we discuss searching a field of correlated noisy observations for signals that manifest themselves as localized “bumps” in the random field. Starting from the Poisson clumping heuristic, we suggest a definition of False Discovery Rate (FDR) when positive results are defined by a clump of observations exceeding a suitable threshold, and we discuss additional analyses designed to identify configurations of true positives.

This is joint research with Benny Yakir and Nancy Zhang.

Keywords: FDR; Scan statistic; Poisson clumping heuristic; Statistical genetics

Biography: David Siegmund is the John D. and Sigrid Banks Professor at Stanford University.

His current research focuses on problems of signal identification in genetics and related molecular biology.