Integrating Multiple Omics Analysis For Microbial Detection,Genome Annotation And Their Interaction With Host

Pathogens continue to threaten public health and life worldwide today. Fast and effective detection will improve our response to global-scale surveillance significantly. At present, microbial detection is based on PCR technology, and thus it’s essential for us to design sensitive and specific PCR primers. In this thesis, based on nucleotide sequences in the public database, a novel workflow is built to design primers for common respiratory infection viruses such as influenza. These primers are conserved to result in robust amplification successfully within species/subtypes, while they also have the ability to discriminate between the species/subtype of interest and closely related species/subtypes.When the pathogen is known, the next step is to study its genomic function and try to understand infection and pathogenic strategies. This will help prevent and treat various diseases caused by pathogens. Therefore, it’s necessary to develop computational methods for accurate genome annotations. Traditional methods including homology search may fail in the cases where genes are evolutionarily distant and do not have functionally known homolog in the databases, or where homologs are hypothetical in the databases. Through protein-protein interactions and the assumption that genes which are associated or interacting are more likely to share function, we develop a novel computational approach to annotate gene function. In the case of identifying virulence factors in bacterial proteomes, the prediction accuracy is up to90%, which is much better than BLAST and machine learning method developed by others. In particular, in the last10years, high throughput sequencing technology reduces the time and economic cost of obtaining full genomes of pathogenic microorganism significantly. This generates a large amount of genome sequences with many regions of unknown function. By introducing the project of Escherichia coli genome sequencing and comparative genomics analysis, the bioinformatics analysis pipeline is illustrated, helping researchers mine information about function and phenotype of genome.Finally, since interspecies interactions play a key role in the host disease caused by pathogens, we analyze virus-host interactions systematically from the perspective of protein domains for the first time. It is shown that viruses preferentially tend to target hub and bottleneck domains in the host cellular interactome, and different types of viruses perturb the host network by common and specific strategies:for example, on one hand, DNA, RNA and Retro-transcribing viruses target common host domains, on the other, they use their own domains to competitively interact with host counterparts. Moreover, the viral infection and genomic variations apply similar ways, e.g., viral targets are enriched in disease-related genes or domains, to alter the local or global properties of human intracellular network, inducing pathological states. Combining systems-level studies of virus-host interactions with genomic approaches will shed light on the molecular mechanisms of human disease.