Selecting differentially expressed genes (DEGs) is one of the most important tasks in microarray applications for studying multi-factor diseases including cancers. However, the small samples typically used in current microarray studies may only partially reflect the widely altered gene expressions in complex diseases, which would introduce low reproducibility of gene lists selected by statistical methods. Here, by analyzing seven cancer datasets, we showed that, in each cancer, a wide range of functional modules have altered gene expressions and thus have high disease classification abilities. The results also showed that seven modules are shared across diverse cancers, suggesting hints about the common mechanisms of cancers. Therefore, instead of relying on a few individual genes whose selection is hardly reproducible in current microarray experiments, we may use functional modules as functional signatures to study core mechanisms of cancers and build robust diagnostic classifiers.
differentially expressed gene, functional module, classification, diverse cancers
Citation:Yao C, Zhang M, Zou J F, et al. Functional modules with disease discrimination abilities for various cancers. Sci China Life Sci, 2011Sci, , 54: 189 -193, doi: 10.1007 With data from microarrays, one of the most important tasks is to identify differentially expressed genes (DEGs) which would be "significant" or "important" enough for the follow-up studies [1][2][3]. However, DEG lists produced from different studies for a cancer are usually highly inconsistent [4,5]. As recently demonstrated by us [6], even when technical measurement variations are small, DEG lists correctly determined by a statistical method such as SAM (Significance Analysis of Microarrays) [1] from different microarray studies for a cancer still tend to be highly inconsistent. Under such situations, selecting DEGs solely based on statistical cut-off criteria (e.g., a false discovery rate (FDR) level) becomes less relevant [7][8][9]. However, functional modules significantly enriched with DEGs tend to be robust to the uncertainty introduced by DEG selection [10,11], as also formally demonstrated by us recently [12]. In this work, we used another approach to identify functional modules with high disease discrimination abilities for cancers. Our results demonstrated that many functional modules have high disease discrimination abilities for the five types of cancers originating in different tissues, indicating that each cancer involves the disruption of various cellular processes and genes in these functions undergo broad expression changes [13,14]. Especially, the modules shared by the five types of cancers also cover a wide range of functions, suggesting the mechanisms shared by various cancers. Obviously, understanding the mechanisms shared by various cancers has important diagnostic implications. For example, the selected modules can be used as functional signatures to build robust cancer diagnostic classifiers [15].