Bioinformatics for medical diagnostics: assessment of microarray data in the context of clinical databases

To identify genes suitable for medical diagnostics microarray data is assessed in the context of clinical databases, which store complex information about the patient phenotype. The wealth of data and lacking standards make it difficult to analyse this kind of data.

We present a workflow for exploratory analysis of microarray data together with clinical data consisting of four steps: definition of clinically meaningful research questions in a masterfile, generation of analysis files, selection and characterization of differentially expressed genes, and estimation of classification accuracy. We applied this workflow to large data sets from the field of cardiology and oncology (n~500 patients). Systematic data management of microarray data and clinical data helps to make results more transparent and comparable.