Predicting the Driver Variants and Mutations in Lung Cancer Genome using Transcriptional Regulation Network

Abstract

Among thousands of potential mutations, identifying and separating cancer driver genes remains a big difficulty. Precise identification of driver genes and mutations is crucial for cancer research and treatment personalization based on accurate patient classification. Many driver mutations within a gene exist at low rates due to inter-tumor genetic variability, making it difficult to identify them from non-driver mutations. Proposed model uses a transcription adjustment network and its data set from the database REGNETWORK. The subject of the paper is to discovery of genes that cause lung cancer with a network approach. To do this, centralization and socialization in graph is used. The degree of centrality, degree of mediocrity, and proximity are considered as parameters in identifying lung cancer gene (with the cancer-causing mutation). Socializing of data is implemented to find genes that are more closely related to each other. Various transcription factors, genes, and their interconnections make create a particular class of biological network called a transcriptional regulatory network. These networks were analyzed to look at how information moves through a biological system and to spot paths that are advantageous for various tasks. Nodes in this network are genes and transcripts, so there are two types of modules in the network, Gene module and transcription factor module. Edges represents physical or regulatory interaction between them Two different algorithms are used to build the network model and comparison is done using Accuracy, F1 Score, Recall, Precision. Using this algorithm, the influential genes (propagation occurs in them) are identified in each community, and finally the total of the influential genes. The results from all communities were predicted as lung cancer genes and evaluated using certain criteria.