Single nucleotide polymorphism--disease relationships: statistical issues for the performance of association studies.

摘要

There is increasing concern about the low reproducibility of findings of genetic association studies in complex diseases. In the present paper, we argue that it might be beneficial to learn from epidemiology and the methodology for clinical trials which address problems also crucial for genetic association studies. We propose a two-step sequential approach in order to (a) allow a study--internal control of reproducibility and (b) stop a study or lab analysis prematurely to save valuable specimen, time and costs. Additionally, minimal information is proposed which should be presented in publications about association studies in order to facilitate reproducibility studies by other authors. Copyright 2002 Elsevier Science B.V.