Gastric Cancer Molecular Typing Based On Bioinformatics Methods

[Background and Objective] The process of gastric cancer is the result of thecombined effects of environmental and genetic factors. In China, the incidence andmortality of gastric carcinoma topped the list of carcinomas, which is a serious threatto public health. Nowadays, surgical resection combined with radiotherapy andchemotherapy is the major treatment for gastric cancer, and the five years survival rateof its early stage after treatment can nearly reach90percent. Hence, early diagnose andtreatments are essential for gastric cancer treatment. Lauren which is a gastric cancerpathology typing method can be used as an independent prognostic indicator for gastriccancer, among which the prognosis of intestinal type is better than diffuse type.Microarray is a high-throughout technology that has been utilized widely in studies ofgastric cancer, through which large datasets of gastric cancer have accumulated. Datamining of these datasets would provide a new way to underline the molecularmechanism for process of gastric cancer. In this study, we collected a set of gastriccancer microarray raw data, and selected same gastric cancer specific biomarkers basedon data mining, machine learning and other bioinformatics methods. Finally, we sortout a set of biomarkers can used for Lauren typing, which would provide new sightsfor gastric cancer molecular typing.[Materials and Methods] Microarray data meta-analysis was firstly performed tosort out differentially expressed genes. We then performed pathway enrichmentanalysis, characteristics selection, and clustering analysis to screen gastric cancerbiomarker genes. And finally, we used qRT-PCR to validate the gene expression levelof these biomarkers. RadViz a visualization method was utilized to perform Laurentyping analysis.[Results] We totally selected117consistently differentially expressed genes.Through pathway enrichment analysis, characteristics selection, and cluster analysis,we then detected9biomarkers, which would accurately identify the gastric cancer tissue samples from normal samples. QRT-PCR validation results showed that AHCY,NUP107and UBE2C would be utilized for Lauren typing analysis.[Conclusion]1) gene expression profiling microarray datasets sample informationretrospective analysis combined with feature selection and hierarchical analysis, abiomarker dataset consisted nine genes (AHCY, NUP107, PSMA7, PSMB2, UBE2C,ALDH1A, ATP4B, CKB and KCNE2), can accurately identify the gastric cancer tissuesamples from normal samples, with a high accuracy about98.09%.2) By qRT-PCR validation, a gastric cancer classifier constructed by AHCY, NUP107and UBE2C can accurately classified the Lauren classification samples.