Identifying Differentially Expressed Genes Based on Differentially Expressed Edges

Identification of differentially expressed (DE) genes under different experimental conditions is an important task in many microarray-based studies. There are many methods developed to detect DE genes based on either fold-change (FC) strategy or statistical test. However, majority of those methods identify DE genes by calculating the expression values of individual genes, without taking interactions between genes into consideration. In this study, we consider the interaction and importance of genes in the network and believe that the edges in the network also contribute a lot to DE genes. Therefore, we propose three new ideas for calculating the expression values of edges by considering mean expression, minimal expression and partial expression, respectively. Those methods were implemented and evaluated on the microarray data and were compared with existing methods. The results show that the proposed edge-based methods can identify more biologically relevant genes and have high computational efficiency. More importantly, the Min-Edge method outperforms the other methods when feasibility and specificity are considered simultaneously.