Toward a combinatorial approach to the sample size problem in multiple-treatment microarray studies


The determination of a sufficient sample size for a microarray experiment with multiple treatments is an important problem in Bioinformatics. While most current approaches have been statistically based, specifically on power calculation, we propose a novel combinatorial approach to studying this problem by relating sample size to the rate of "contractibility" of gene responses represented by transitive directed acyclic graphs (tDAG). Soundness of the proposed method was argued theoretically and justified with results of carefully designed experiments on both real and synthetic data. As a result, we propose a computational procedure that suggests a sufficient sample size of any microarray experiment with multiple treatments.

Publication Title

Proceedings of the 2008 International Conference on Bioinformatics and Computational Biology, BIOCOMP 2008

This document is currently not available here.