Detection And Analysis Of Gene Tissue Specificity Based On High Throughout Expression Data

In recent years,with the rapid development of high throughout technology,a large amount of bioinformatics data has been produced,in which there are a lot of important information with regard to physiology and medicine.It is of great significance to study the function study of cells and tissues by mining tissue specificity information from the massive biological data.In this thesis,based on high throughout expression data,the main research works focus on gene-specific identification index assessment,housekeeping genes(HK genes)and tissue-specific genes(TS genes)classification,gene-specific analysis.By surveying papers,we built a standard database of HK genes and TS genes to compare and analysis the tissue-specific identification index.According to the results,the machine learning was the best method to identify gene as HK gene or TS gene with the RNA-seq data sets.By comparing the nearest neighbor classifier,the Bayesian classifier,the decision tree classifier and artificial neural network classifier,the SVM with RBF kernel had the best classification performance.After 10-fold cross-validation,we chose the best SVM prediction model which identified 8395 HK genes and 5162 TS genes.By analyzing the expression pattern,the HK genes and TS genes' expression pattern was closely related to the specific of tissues.Then we picked out 2557 constant expression genes from identified HK genes.UBC and GADPH were considered the prefect reference genes.We further confirmed and enriched the physiological functions of HK genes and TS genes through genes' GO annotation and KEGG pathway analysis,which laid a foundation for a comprehensive understanding of gene specific.