Abstrakt

Using differential nonlinear gene co-expression network analysis for identification gastric cancer related genes

Hesham Abdulatef Mohammed Al-Bukhaiti, Jiawei Luo

Gastric cancer is one of the most common malignancies and ranks the second highest mortality in the world today. With this great progress, understanding of tumors is still a wide area to study. We utilized the distance correlation as the measurement of the relevance to construct the gene co-expression networks in both gastric cancer and normal tissue datasets respectively, also to define the complex interactions among genes with nonlinear property preserved and examine from the genes that have important roles in the gastric cancer formation. The genes which are extant in the module of normalassociated network but missing in the module of cancer-associated network are selected for further study and we call these genes stray genes. The results show the stray genes are enriched in up-regulation and the connectivity of all the genes appear the circumstance of loss in the gastric cancer network, especially to the stray genes. These results indicate that the activation of these stray genes may play important roles in the gastric cancer. We use PANTHER for Gene Ontology (GO) analysis and the results show that these stray genes are enriched in some biological processes, including cell cycle, chromosome segregation, DNA replication and p53 pathway and others. These results prove the effectiveness of our method for cancer-related genes identification and these cancer-related genes can be selected for further analysis.