Biomarker Prediction For Pancreatic Cancer And Liver Cancer Based On Microarray

Cancer is also called malignant neoplasm, which is a kind of disease lacking ofreproduction control. Cancer cells not only grow and reproduce without control, but alsoinvade into other organs nearby. Some of them even arrive at organs more far through lymphand blood system. Different kinds of cancers are with different malignant degree. For example,pancreatic cancer five-year-survival rate is about6%, meanwhile prostate cancerfive-year-survival rate is almost100%. However cancers diagnosed in the same organ areeven with different malignant degree. Taking an example, small cell lung cancer is moremalignant than non-small cell lung cancer. Based on previous research, the reason why somekinds of cancers are with high death rate is on one hand their malignant degree, and on theother hand lacking of effective diagnosis methods which lead to treatment in time. Lots ofcancer patients are on late state when diagnosed.It’s significant to research on cancer gene expression difference for uncovering thepathology of cancers and application on clinical diagnose. Previous research on cancer geneexpression difference is based on analysis of a gene set in a single cancer. Those methodsneglect the information on cancers differences and commons. Utilizing those neglectedinformation will provide an enlightenment for uncover the secret of the cancer pathology orthe reason leads certain cancers with high death rate. And the related genes could be thecandidate biomarkers of high death rate cancer. It is of significance for early diagnosis.In this paper, we acquire cancer gene biomarkers with high death rate by comparing highdeath rate cancer group, which contain pancreatic cancer and liver cancer and low death ratecancer group, which contain breast cancer and prostate cancer. Common selecting genesmethods are based on single gene selection. Those methods neglect the complication ofbiology system. The genes in a biology system are with strong relationship between eachother. Thus we use the bicluster algorithm to avoid those lacks above. And we find the bestparameters of the bicluster using genes filtered in roughly.Based on those parameters, we analysis genes expression difference from these four typesof cancers and we acquire240high death rate gene biomarkers ranked by frequency. The setof genes is with biology meaning. For example, one pathway acquired is about focal adhesion.It’s a kind of macromolecular for conducting signals and controlling signals. It’s contractedwith extracellular matrix, moreover collect and transmit signals at the spot of the integrin likea signal harbor. There is strong relationship between interaction of extracellular matrix and cancers. And it could be the reason that leads the high death rate. The genes on top are alsorelated to these high death rate cancers. Taking a example, UBE2S is related enzyme for celldivision. The malignant degree is related to the frequency of the cell division. The rapiddivision could accelerate the mutation of all the genes in the cell which means to acceleratethe process of cell evolution. It could lead the cancer cells more accustomed to theenvironment around by nature selection.Finally, the high death cancer group could be classified from the low death cancer groupby utilizing the gene feature set. These genes could be the potential gene biomarkers for highdeath rate cancer.