AMultivariate Approach to Integrating Datasets using made4 and ade4

The public microarray repositories, ArrayExpress and the GeneExpression Omnibus (GEO), now contain over 100,000microarray gene expression profiles (Table 1). This is a considerable data resource. However the average number of arrays per study is only between 30 and 40 (Table 1). Given that the number of features (genes) on microarrays now exceeds 50,000, this presents a considerable dimensionality problem. Low case to feature ratio is likely to remain an issue, as cost and availability of biomaterial, such as biopsy tissue, are often limiting. As a result, meta-analysis or merging data from multiple studies is attractive.