The high cost and difficulty of obtaining high-quality mRNA from primary tissue has led many microarray studies to be conducted based on the analysis of single sample-replicates. The purpose of our study was to quantify the impact of this practice on the quality and reproducibility of reported results by exploiting the multiple-array-per-chip design of the Illumina BeadChip platform. Intra-experiment technical variation in Human WG-8 chips was assessed using 18 repeat hybridisations of universal human reference RNA (UHRR) and 64 duplicate hybridisations of primary breast tumour samples from a clinical study.
A clear batch-specific bias was detected in the measured expressions of both the UHRR and clinical samples. This bias was found to persist following standard microarray normalisation techniques and, following analyses for differential expression using two popular methods, it was observed to have a substantial effect on the reported gene-profiles. In this seminar I will describe the design, analysis, and results of this particular study and provide a brief coverage of batch-processing effects observed in data obtained using other array platforms. I will discuss the mounting evidence in favour of performing specialised batch-corrections on microarray data obtained from a single experiment and also when integrating different datasets at the expression level. Finally, I will introduce valuable batch-correction algorithms that minimise these effects and discuss implications for the design of future experiments employing microarray technologies.