Building a Cancer Progress Model Based on PPI Network and Gene Expression Changes and Its Applications for Classification of Cancer Subtypes

he different gene expression profiles always exist in different cancers. Even if the cancers are the same in the same stage, the differences still exist in the gene expressions, not to mention the different cancers on the same site. To distinguish different cancers on the same site is very hard. In the early cases, differences need to be distinguished by doctors. Then further differences could be found by pathological experiments. For example, the brain tumor can be classified by tumor origins, cell type origin or the tumor site etc. With advanced bioinformatics and accumulation of gene expressions, researchers could try to use these data to cluster cancer subtypes. However, the operation of biopathway is more matched the mechanism of cancer, the application of biopathway for the clustering of cancer subtype is not sufficient. In this study, we collected gene expression data of healthy and four myelodysplastic syndrome subtypes and applied a method that integrated protein-protein interaction and gene expression data to identify different patterns of disease subtypes. We hope it is efficient for the clustering of cancer subtype.