The prognostic role of genes with skewed expression distribution in lung adenocarcinoma

Abstract

Many studies assumed gene expression to be normally distributed. However, some were found to have left-skewed distribution, while others have right-skewed distribution. Here, we investigated the gene expression distribution of five lung adenocarcinoma data sets. We assumed that samples in the tail and non-tail of a skewed distribution were drawn from different populations with different survival outcomes. To investigate this hypothesis, skewed genes were detected to build a tail indicator matrix comprising of binary values. Survival analysis revealed that patients with more skewed genes in their tails had worse survival. Hierarchical clustering of the tail indicator matrices discovered a gene set with similar tail configurations for either left or right skewed genes. The two gene sets divided patients into three groups with different survivals. In conclusion, there is a direct association between genes with skewed distribution and the prognosis of lung adenocarcinoma patients.