Mining topological structures of protein-protein interaction networks for human brain-specific genes.

Compared to other placental mammals, humans have unique thinking and cognitive abilities because of their developed cerebral cortex composed of billions of neurons and synaptic connections. As the primary effectors of the mechanisms of life, proteins and their interactions form the basis of cellular and molecular functions in the living body. In this paper, we developed a pipeline for mining topological structures, identifying functional modules, and analyzing their functions from publically available datasets. A human brain-specific protein-protein interaction network with 1482 nodes and 3105 edges was built using a MapReduce based shortest path algorithm. Within this, 7 functional cliques were identified using a network clustering method, 98 hub proteins were obtained by the calculation of betweenness and connectivity, and 5 closest relationship to clique connector proteins were recognized by the combination scores of topological distance and gene ontology similarity. Furthermore, we discovered functional modules interacting with TP53 protein, which involves several fragmented research study conclusions and might be an important clue for further in vivo or in silico experiments to confirm these associations.

Analysis of differentially expressed genes in malignant biliary strictures.

Microarray data were collected from bile duct samples from subjects with malignant biliary strictures by endoscopic retrograde cholangiopancreatography to screen for key genes associated with this disease. A predicted interaction network was constructed for these genes to interpret their functions. The gene expression dataset GSE34166 (10 samples: 6 malignant and 4 benign control samples) was downloaded from the Gene Expression Omnibus database. R package scripts were used to process the data and screen for differentially expressed genes. Genes identified were uploaded to the analysis tool String 8.3 to generate a gene interaction network. A hub gene was identified by calculating the node degree. The interaction network of the hub gene with other genes in the human genome was constructed and screened (score >0.9), and pathway-enrichment analysis was performed to elucidate the hub gene function. In total, 377 differentially expressed genes were identified and a network comprising 209 pairs of interactions was constructed. The most critical hub gene was identified as GSTA1, and a GSTA1-based interaction network was constructed consisting of 25 genes (containing the differentially expressed gene GSTA3). The cytochrome P450 (CYP450)-metabolic pathway displayed the most significant enrichment. Additionally, 4 transcription factors and their binding sites were also identified. In conclusion, we have identified the differentially expressed genes GSTA1 (a hub gene) and GSTA3, which may cause abnormal gene expression and tumorigenesis through CYP450-metabolic pathways. The transcription factors and their binding sites in the promoter of the hub gene provide potential directions for future drug design.