Validation of the gene stability measure and thegeometric averaging of carefully selected control genes for normalization. (a) Validation of the gene stability measure. The average gene-specific variation (determined as coefficient of variation, in percent) for the three control genes with the smallest variation within each tissue panel after normalization with three different factors calculated as the geometric mean of the three control genes with the lowest (NF3(1-3)), highest (NF3(8-10)) and intermediate (NF3(6-8)) gene stability values (as determined by geNorm). NB, neuroblastoma; POOL, normal pooled tissues; LEU, leukocytes; BM, bone marrow; FIB, fibroblasts. (b) Geometric averaging. Comparison of frequently applied microarray scaling factors and the proposed RT-PCR normalization factor based on the geometric mean of selected control genes (NF5, geometric mean of the five control genes with the lowest M value; NFM < 0.7, geometric mean of control genes with M < 0.7; see Results), calculated for eight hybridizations from publicly available microarray data .