Identification And Functional Inference For Tumor-associated Long Non-coding RNA

Cancer is the number one killer of human health, scientists and doctors around theworld had been devoted their lives to studying and researching the reason and mechanismof cancer for a long time, and they try to find efficient methods to prevent, diagnose,monitor and cure cancer. In the life sciences, the analysis of DNA’s, RNA’s and proteins’functions has always been very important work. Since the beginning of2006, the newgeneration of sequencing technology greatly reduced the cost of genome sequencing, alarge numbers of countries had launched the Cancer Genome Project, produced a hugenumbers of cancer genome data (like TCGA), providing more and more gene mutations andexpression profiles of cancer, it provided more comprehensive data to support the researchof cancer. As the rapidly growth of biological data, it was becoming impractical to rely onmanual analysis, so people must use computer science and technology to analyze andprocess large amounts of biological data.A noncoding RNA is a functional RNA molecule that is not translated into a protein.Long noncoding RNA (long ncRNA, lncRNA) is a non-protein coding transcript longerthan200nucleotides. As they do not code proteins, people thought they were meaningless.But as people gradually researching the function of noncoding genes, through theobservation of various body’s physiological and pathological processes, they discoveredmore and more biological functions of noncoding genes. Noncoding RNAs are also closelyrelated to the occurrences and development of diseases. They have differential expressionin different tissues, healthy or not human bodies, and even in the bodies of old people or not.As a result, it’s necessary to study the function of long noncoding RNA. And we wouldresearch the differential expression and function of the long noncoding RNA as a system.As a large number of lncRNAs have been identified, researchers found that someprobes of exon arrays in GEO were mismatched mRNAs, which actually corresponding tolncRNAs. Compared with expensive RNA-seq technology and specially designed lncRNAchips, there are a lot of cancer related exon array chip data, we can quickly infer theexpressions of the lncRNAs in different tumors and the coexpression between lncRNAs andproteins. It provides plentiful data resources for the study of tumor-associated lncRNAs’ different expression and functional inference on the system level.The main work of this paper is in these steps. Download a large number of humantumor exon array data on Human Genome U133Plus2Array platform in GEO database.There are three categories in these data, one of them contains glioblastoma brain tumors inchildren and adults data set, the next one contains16different cancers exon array data set,the last contains a sample set of four-stage colon cancer. Next we re-annotated the probeswhich actually corresponding to lncRNAs and obtained the expression of part of thelncRNAs and coding genes. So we could calculate the fold change and p-value of theexpressional data in disease and control groups. The genes whose fold change is greaterthan2or less than0.5and p-value is less than0.05are considered as significant differentialexpressing ones. And then we calculate Pearson product-moment correlation coefficient andSpearman’s rank correlation coefficient of these differential expressing genes, resulting inthe co-expression network associated with lncRNA. Next we sent the coding genes in thenetwork to DAVID, a Database for Annotation, Visualization and Integrated Discovery, itwould give you their GO biological process and KEGG pathway enrichment. From this, wecan infer the biological functions of the tumor-associated lncRNAs and get a newbreakthrough for the mechanism researchment of tumor.