The most effective QC/QA process starts with a good experimental design. Our experience with GENEVA and other projects has led to the following recommendations. Adherence to sound epidemiological principles in the recruitment of subjects for case-control studies is a crucial first step [Zondervan and Cardon 2007]. Subsequently, it is critical to avoid association between case-control status (or other phenotypes) and any variables that may affect genotyping quality. Variables that may affect DNA sample quality include tissue collection date, storage and shipping conditions, tissue type, DNA extraction method, extraction batch, and study site. Factors that may affect genotyping process quality are reagent batches, instrumentation and processing batches (such as plate effects). Therefore, to avoid confounding, we recommend balancing case-control status across experimental factors and randomizing the order of processing and the plate positions of samples. We also recommend the use of at least five duplicate sample pairs to assess the overall accuracy of genotyping and at least 30 pairs for SNP filtering. The duplicates should be selected to represent the overall quality and tissue source of the study samples. Family trios
