Background: The maturing of gene expression microarray science and interest in the use of microarray-based applications for clinical and diagnostic applications calls for objective measures of quality. This manuscript presents a retrospective study comparing Affymetrix microarray platform data from both a standard mixture of external spike-in controls and endogenously-present internal controls with array quality assessment using full-array metrics. Results: A layered PCA modeling methodology of the individual classes of controls (spike-in hybridization, spike-in polyA+, internal RNA degradation, endogenous or "housekeeping genes") is presented for the evaluation of microarray data quality over the various stages within the experimental protocol (e.g., hybridization, RNA amplification). The controls available in these microarray experiments provide a continuum of data quality information: external spike-in hybridization and RNA labeling controls provide information related to with both assay and hybridization performance whereas internal endogenous controls provide primarily quality information inherent to sample RNA integrity. Overall, we find that the variance present in both the external and internal controls carry with them the critical information about technical performance, consistent with whole-array quality assessment as provided by the wealth of quality assurance/quality control (QA/QC) metrics described in the MiMiR database. Conclusions: This work provides validation for the use and analysis of both external and internal RNA control data to assess the technical quality of microarray experiments. Consistency of the information carried by the both internal and external controls and the conventional quality measures will help validate the future use of rationally-designed control standards for routine performance monitoring of multiplexed measurement platforms.
Citation: BMC Genomics
Pub Type: Journals
microarray, quality assessment, quality control, spike-in controls, hybridization, RNA amplification, exogenous and endogenous controls