Uncovering The Neuropsychiatric Disorder-related Pathways And Their Function Based On Protein-protein Interaction Network

Neuropsychiatric disorders are complex diseases threatening human health seriously.Systematic exploration of the pathogenesis of neuropsychiatric disorders is helpful in improving the clinical diagnosis and treatment measures and patients’ life quality.With the development of the genomics and related techniques,more and more gene,protein or other molecules associated with various neuropsychiatric disorders have been identified via different approaches.However,how to uncover the meaningful molecular mechanism(i.e.,dysfunctional pathway and their cooperation relationships)underlying the astronomical amount of genetic data accurately is still one of the major challenges of biomedical research including the neurology area.Meanwhile,with the the existence of extensive molecular interactions,the widely adopted research models focusing on single or a few molecules have been more and more constrained in uncovering the molecular mechanism of complex diseases.Consistently,in order to understand the molecular mechanisms underlying a disease,approaches that tacking the problem from systems biology aspect become more and more necessary and important.In this thesis,based on Protein-Protein Interaction Network(PPIN),we used mathematical modes of statistics and graph theory to address some of the main challenges in pathway analysis including pathway identification and relationship measurement among pathways.Then,we provided a more comprehensive workflow for pathway analysis.Meanwhile,combined with the characteristics and development of the researches in neuropsychiatric disorders,we performed systematic pathway analysis on candidate genes related various neuropsychiatric diseases aiming to understand these disorders as well as their co-mobidity deeply.The main points of the work can be summarized as follow:(1)We proposed a method to assign a function weighting score to each candidate gene based on their correlation with disease.Further evaluation of function annotation indicated the function weighting score scheme was consistent with available evidence.Then,by incorporating the function weighting scores of candidate genes into conventional pathway analysis,we developed a novel pathway identification approach.According to the non-equivalent function score,more important genes had a bigger role in pathway analysis and might uncover the more comprehensive dysfunctional pathways,especially for disease-related genes lacking in numbers.Meanwhile,this approach could reduce the interference from unimportant genes in pathway identification.We further identified the dysfunction pathways participated in nicotine dependence and demonstrated that neurodevelopment-and metabolism-related pathways had key role in nicotine dependence.Compared to conventional over-representation based pathway analysis,the novel pathway detection method exhibited improved discriminative power and detected some novel pathways potentially underlying nicotine dependence.(2)In biological system,a pathway is a series of actions or interactions among genes or genes products that leads to the generation of molecular network with a certain product or a change in a cell.Besides,according to the disease module hypothesis,the candidate genes associated with a disease segregate in the same neighborhood of the PPIN and have a tendarcy to merge into disease-related subnetwork.Meanwhile,the “isolated” genes are very probably “noisy” genes.Here,we proposed an approach to identify the dysfunctional pathways by calculating the distribution feature of pathway-related genes in disease’s sub-network,especially for disease-related genes with large quantities.By constructing the disease-related subnetwork from PPIN,we could remove the disturbing genes from the disease-related gene list and mitigate their impact on pathway detection.Via calculating two features including largest connected component and shortest path distance among pathwayrelated genes distributed in disease’s sub-network,we identified dysfunctional pathways involved in disease.Analyzing several gene lists demonstrated that our proposed pathway detection approach could provide high-efficiency performance and best cleaning effect.Finally,we performed pathway analysis on candidate genes associated with multiple sclerosis(MS)and identified key dysfunctional pathways.Among these pathways,apoptosis-neuroimmune-,oxidative stress-related pathways were the hub pathways for MS.(3)In biological system,the relationships among pathways can be cooperative,compensatory or alternative and detecting the relationships among pathways is essential for understanding the association between pathway and biological processes.Identifying dysfunctional pathways associated with specific diseases represents only the first step of a systematic program toward understanding the mechanism of complex diseases.However,currently most pathway analysis approaches only provide several “isolated” pathways potentially participated in disease and ignoring the interplay of these pathways.Focusing on the complex relationship among pathways,we proposed a method to measure the relationship between pathways based on their distribution in the human PPI network.By representing each pathway as a gene module in the PPI network,a distance was calculated to measure the closeness of two pathways.For the pathways in the KEGG database,a total of 2143 pathway pairs with close connections were identified.Additional evaluations indicated the pathway relationship built via such approach was consistent with available evidence.Further,based on the genes and pathways potentially associated with the pathogenesis of Parkinson’s disease(PD),we analyzed the pathway relationship and identified the major pathways related to this disorder via the new method.Also,by analyzing the pathway interaction network constructed by the identified major pathways,we explored the potential pathway targets that may be important in the etiology and development of PD and identified 4 pathway modules including neurodevelopment-,oxidative stress-,infectious disease-and immune-related pathway modules.(4)Clinically,co-morbidity of neuropsychiatric disorders is a common issue.Here,we explored the molecular mechanism underlying the high comorbidity of tobacco smoking and schizophrenia systematically.We built a schematic molecular network for nicotine addiction and schizophrenia based on the results of pathway and network analysis,providing a systematic and straightforward view to understand the relationship between these two disorders.Besides,we identified 11 novel candidate genes potentially associated with the two diseases and further literature retrieval found supporting literature for most of these genes.We also analyzed the mechanism underlying the influence of bio-behavioral factors on ovarian cancer.Results demonstrated that infectious events and immune diseases might trigger the biobehavioral factors impact on ovarian cancer and inflammatory factors might be the key regulators.According to pathway network analysis,we identified 3 hub pathways might play important roles in this disease.Meanwhile,5 pathways were predicted as potential pathogenetic mechanism involved in the ovarian cancer with high biobehavioral factor and neurotrophin signaling pathway deserved in-depth explored.In summary,from pathway detection to pathways’ relationship measurement,we provided a relatively complete workflow of pathway analysis based on PPIN.On this basis,we performed systematically pathway analysis by appropriate methods on several neuropsychiatric disorders as well as their co-morbidity.Our proposed methods have important value in uncovering the mechanism underlying these diseases.