Research On Relevant Computational Problems In Virus-host Protein Interaction Network Analysis

Viral infectious diseases are endangering human health everyday. Throughout history there have been a series of pandemics, such as smallpox and tuberculosis. Most recent ones, by SARS, HIV and H1N1 have created tremendous social and economic disaster. Bioterrorism may also manipulate these potential weapons. Significant achievement has been made antiviral drug studies during the past two decades. Most conventional antiviral drugs, however,target proteins virus-encoded, that allow virus escape drug pressure by rapid evolution and resistant adaptation in the viral genomeIn one word, the diversity and enormous evolutionary potential of the virus is the major challenge in the broad-spectrum antiviral drug discovery.Although resistance and its replication requires virus-derived factors viruses, as obligate intracellular parasites, totally depends on host cellular functions, It is plausible that an agent?reagent/chemicals will be identified that targets one or more critical host proteins which is essential for pathogen entry, survival, and replication. Strategies of targeting host cellular factors are advantageous because, in principle, they are far less susceptible to escape mutation in response to selection pressure than those of targeting virus directly. Thus, from an evolutionary perspective, targeting host cell factors may provide a long-lasting antiviral effect. But the similarity of host-directed and virus-directed processes makes it difficult to identify the host-directed targets with sufficient specificity. Most potential host-directed drug targets are identified by genetic association or from existing druggable genes or protein targets. Comparing to the expensive and time-consuming genetic screens, which based on biochemical tests and cellular assays, Genomics and proteomics is beginning to provide real alternatives to these traditional methods for drug target identification. It combines genome-wide computational biology, genomics, proteomics, and traditional forward and reverse genetics have identified host-virus interactions and host functions critical for the establishment of viral infection. In this paper, we present research on relevant computational problems in virus-host protein interactions network analysis that may be critical factors for host-directed drug targets discovery by the systems biology approach for integrating high-throughput experimental data and large scale protein interaction map.Identification of the host-directed targets requires a detailed knowledge of host-pathogen interactions. First, we integrate human–human and human–virus protein interactions for 10 virus strains from currently available public databases. And topological structure analyses are performed on these protein interaction networks to identify topologically important proteins, which are more crucial in the infection process.Further network analysis needs to integrate dynamic gene expression profile and protein interaction networks, which requires the comparison, and integration of data sets obtained with different microarray platforms. The gene expression profiles based on different experimental protocols for manufacturing the microarrays, hybridization, and image analysis make comparison of the data across platforms difficult. it is impossible to compare profiles from different microarray platforms due to many obscuring sources of variation. Here, we present a highly verifiable inter-chip normalization method using genetic optimization algorithm, which depends only on a small number of housekeeping gene, and therefore facilitates experiment validation without any accuracy lost. Compare to invariant set normalization, only a small optimal set of housekeeping genes are selected as normalization markers, which makes the experimental verification applicable for accuracy and reliability of inter-chip normalization.Protein interaction networks are highly dynamic and responsive to signals from the environment. Traditional analysis of gene expression data focused on identifying differentially expressed and co-expressed genes, which did not take known interactions into consideration. In recent years, many methods have been developed to identify active subnetwork by integrating protein interaction networks with gene expression profiles. Current approaches failed to take full account of both difference and correlation in gene expression that may lead to false positive results. A new algorithm is proposed for identifying active subnetwork by considering both difference and correlation of gene expression profile. The algorithm is employed in the process of gene expression profiles of human immunodeficiency virus infection. The results showed that the algorithm can identify the active subnetwork that has extremely high biologically functional connectivity with human immunodeficiency virus, and effectively avoid the bias introduced by considering differences of gene expression profiles only, i.e., genes less differentially expressed are also included due to high correlations in gene expression.Many applications require a similarity measure between biological networks, such as searching and clustering of protein interaction network with similar biological function. Proteins may play different roles, depending on different cell context. The topology of protein interaction network can be used to estimate cellular context information of proteins. Here, we propose a new local similarity index based on neighbours'functional similarity, and then presents a functional similarity measure method for protein interaction networks, which is based on the globally optimal matching between the proteins of two networks. The clustering analyses of KEGG pathway database and viral active networks show that the outcomes of our method are shown to be consistent with human perspectives. The fluctuation in protein concentration can be described conceptually as a balance between positive and negative factors: accumulation due to gene expression and protein synthesis , and dilution due to an increase in protein degradation. The response of biological system to internal and external stimuli is always accompanied by some flux of protein concentration. The concentration changes in the protein interaction network can be quantitatively described by using the law of mass action. Here, we present a computional estimate method for the decay factor in perturbation propagation. The decay factor can be used to quantify the speed of the decay intensity while concentration changes cascade through the network. The results on the yeast protein interaction network show that the decay factor is negatively correlated with the initial disturbance intensity, which reflects the network is robust against external disturbances. Then this method is applied to identify the proteins vital for those viruses by achieve the optimum coverage with specific disturbance using integer programming model. The results show that this method can identify the key proteins which are essential for viral infection.