Abstract
Prostate cancer is currently associated with higher morbidity and mortality in men in the United States and Western Europe, so it is important to identify genes that regulate prostate cancer. The high-dimension gene expression profile impedes the discovery of biclusters which are of great significance to the identification of the basic cellular processes controlled by multiple genes and the identification of large-scale unknown effects hidden in the data. We applied the biclustering method MCbiclust to explore large biclusters in the TCGA cohort through a large number of iterations. Two biclusters were found with the highest silhouette coefficient value. The expression patterns of one bicluster are highly similar to those found by the gene expression profile of the known androgen-regulated genes. Further gene set enrichment revealed that mitochondrial function-related genes were negatively correlated with AR regulation-related genes. Then, we performed differential analysis, AR binding site analysis, and survival analysis on the core genes with high phenotypic contribution. Among the core genes, NDUFA10 showed a low expression value in cancer patients across different expression profiles, while NDUFV2 showed a high expression value in cancer patients. Survival analysis of NDUFA10 and NDUFV2 demonstrated that both genes were unfavorable prognostic markers.
Highlights
IntroductionThe principle of the biclustering algorithm is to select a subset of rows and columns from a data set and utilize specific measures to maximize the quality of a bicluster, which was first applied to gene expression by Cheng and Church [3]
Prostate cancer has a very high morbidity and mortality rate in men in the United States and Western Europe
We focused on the transcription of differentially expressed genes in the process of androgen receptor (AR) regulation by using ChIP-seq, and screened out six genes that have AR binding sites (SAMM50, NDUFA10, SDHAF4, OXA1L, NDUFS5, and NDUFV2)
Summary
The principle of the biclustering algorithm is to select a subset of rows and columns from a data set and utilize specific measures to maximize the quality of a bicluster, which was first applied to gene expression by Cheng and Church [3]. Based on the gene expression profile, the coexpression analysis method was used to detect gene modules related to prostate cancer. A clustering method is needed to identify the coexpressed gene modules. A huge number of existing biclustering algorithms involve disparate quality measures and search heuristics to explore gene modules [5]. To serve as a measure of quality, traditional biclustering methods detect biologically related biclusters but only find small rather than large gene coexpression modules [9]. Most existing algorithms of biclustering perform well in finding various tiny biclusters involving relatively few genes
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
