Identify differentially expressed genes with large background samples

To identify differentially expressed genes related to diseases is important but challenging. The challenges include the inherent noisy nature of the collected data, as well as the imbalance between the very large number of genes and the relatively small number of collected study samples. To address some of these challenges, here we implemented the method of AUCg (Area Under the Curve gene ranking). The novelty of the implementation of AUCg is that it not only utilises the study samples information but also makes good use of the large amount of publicly available gene expression samples as “background”. We applied AUCg to a private dataset of 217 multiple myeloma samples, compared to 36,754 publicly available gene expression samples. The analysis identified genes that could be potentially unique to multiple myeloma. The AUCg gene ranking method can be applied for studying many other cancers and human diseases, taking advantage of large publicly available data.