T-hücresi akut lenfoblastik lösemi için birbirine bağlı marker tanımlanması

T-cell acute lymphoblastic leukemia (T-ALL) is a very complex disease, resulting from proliferation of differentially arrested immature T-cells. The molecular mechanisms and the genes involved in the cause of T-ALL remain largely undefined. In this study, we found biomarkers to differentiate individuals with T-ALL from the non-leukemia/healthy ones, to discover markers that are not differential themselves but interconnect with highly differentially expressed genes, and to have a network-based view of T-ALL. Instead of applying only expression-based differential gene analysis and obtaining hundreds of candidate disease-causing genes, we integrated gene expression data of T-ALL and healthy samples with the human protein-protein interaction data in order to discover diagnostic biomarkers not as individual genes but as subnetworks. By using a network-based approach, we have identified 19 significant subnetworks, containing 102 genes (out of 409 genes). A given subnetwork contains up-to 12 genes. The classification/prediction accuracies of subnetworks are considerably high, as high as 98%. Some genes in the subnetworks were already known to be associated with T-ALL, but we found new ones to be involved in T-ALL development. The subnetworks were rich in transcription factors whose ectopic activation is known to be one of the reasons behind T-ALL. The Zinc-binding proteins are also abundant in subnetworks. Zinc levels are low in ALL-patients. Zinc supplement given to a T-ALL patient may increase the efficiency of chemotherapy. We recovered 6 tyrosine kinases which have important roles in T-cell survival, proliferation, and immune response. These important genes in our subnetworks may serve as an alternative to the traditional biomarkers used for the diagnosis of T-ALL. The aim of this study is also to help investigators to highlight potential disease gene candidates for further experimental validation.We also applied a typical hierarchical clustering method to most differential 100 and 200 genes between T-ALL and healthy samples. As opposed to the presumption that most differential 100 or 200 genes would classify the diseased samples better, our subnetworks achieved the same or, in some cases, higher classification accuracies. Doing the same/better job with 10 genes in a subnetwork, instead of 100 or 200 genes might be regarded as an accomplishment. In short, network-based classification techniques help us to identify biologically more meaningful subnetworks than expression-based techniques which return thousands of differential genes.