Construction And Analysis Of Diagnostic Network Biomarker For Leukemia

Leukemia is a common malignant tumor with multiple subtypes. Besides the high mortality rate, its incidence is increased these years. Accumulating researches have been reported on detecting single diagnostic biomarkers for leukemia. However, such individual biomarkers are not powerful enough due to the heterogeneity and complexity of this disease. On the other hand, leukemia often originates from disorders of molecules and their interactions at different biological levels. Recently, it is reported that network biomarkers are thought to be more robust compared with individual molecule biomarkers. Furthermore, the biomarkers based on networks are considered to outperform individual molecules in disease characterization.In this study, we presented an integrative analysis that aimed at identifying network modules as potential biomarkers for the diagnosis of leukemia. At first, we integrated the interactions among proteins from the Protein Interaction Network Analysis(PINA) database and leukemia associated genes from GeneGo to construct the leukemia-specific protein-protein interaction network. Secondly the new neural network was integrated with six groups of gene expression profiles for further analysis. Next we mapped p-values in different experimental groups to each gene as the weight of this network. And then, the six networks with different weights were screened through multiple iterations and combinations to identify active modules related to leukemia. Thirdly, we applied overlapping analysis to the active modules and eventually a network with 97 genes and 399 interactions was gotten which could be the candidate biomarkers of leukemia. Finally, the candidate network biomarker was evaluated for the diagnosing performance. Functional enrichment analysis revealed that the network biomarker was mainly enriched in the cancer related pathways. This could demonstrate that the network-based biomarker was associated with the pathogenesis of different types of cancer including leukemia. The result of ROC analysis showed that the network biomarker could discriminate leukemia samples from the normal controls more precisely compared with the reported single biomarkers. We thought that the construction of network biomarker provides an effective approach for the early diagnosis of leukemia and helps understand the molecular mechanisms of leukemia as well.This study provides a novel diagnostic strategy for leukemia from the systematic perspective, which is valuable for both theoretical investigations and clinical practice.