A Research Of Data Mining Strategy For Facing Omics Of Brain Disease In Rat Models

With rapid development of big data acquisition and detection technology,biomedical research has entered omics era,transforming medical data accumulation to medical knowledge mining.The corresponding detail includes more valuable information provision for medical era through precision research mode.Objectively speaking,to achieve this aim,it is essential to apply,improve and develop analysis methods of data mining to obtain and distinguish valuable key knowledge information.It is evident that the analysis of data mining provides methodology basis for medical omics research.This analysis is a key technology of precision medicine under the big data background that has been broadly focused.At present,traditional medical research methods are far from clarifying the occurrence and development mechanism of complex diseases,which makes it difficult to develop effective treatment methods.The brain is the most precise and sensitive organ worldwide,refers to complicated pathologic processes,results to occurrence and development mechanisms that can not be elucidated via traditional medical research methods,and it is hard to find effective therapies.In this regard,this thesis aims to discuss the data mining strategies and methods applied to medical omics research.By integrating,optimizing and innovating data mining analysis methods,it performs omics research(metabolomics and proteomics)on experimental samples for rat models of brain diseases(traumatic brain injury and ischemic stroke)to discover new information characteristics of cerebral pathologic physiology for novel biological significances.It provides a new perspective for exploring the biological significance of omics research.The main contents and novelties are as follows:(1)The plasma of rats with traumatic brain injury model was extracted to obtain metabonomic experimental data.With the help of data mining analysis platform,the identification and classification model of traumatic brain injury diseases was established through data mining strategies and methods such as multivariate statistics,variable selection,enrichment analysis and univariate analysis.The strategy Analyze and verify the statistically significant key information of biological metabolism related to disease characteristics.Based on this strategy,the mining results reveal new evidence of metabolic mechanisms involved in the progression of traumatic brain injury.(2)The brain tissue of ischemic stroke model rats was extracted to obtain quantitative proteomic experimental data.Through data mining strategies and methods such as significant difference judgment,ontology,protein network analysis and enrichment analysis,the statistically significant expressed protein information related to the disease was found.It also reveals the interaction between related proteins and their biological function interpretation,explores and verifies the new proteomic evidence basis for the pathophysiological mechanism of ischemic stroke.(3)A new vissa-plsda algorithm is constructed based on the plasma metabolomics data of the experimental model of traumatic brain injury rats to realize the optimal configuration of variable selection.Through data mining strategies and methods such as variable selection,multivariate statistics and enrichment analysis,this strategy captures the metabolite information related to the pathophysiological process of the disease that cannot be revealed by traditional methods.It establishes a more accurate and reliable discrimination model,improves the efficiency of diagnosing complex diseases,reveals the biological significance of the pathophysiological mechanism of traumatic brain injury,and analyzes as well as verifies the role of metabolites in the process of traumatic brain injury.(4)An analytical model data mining analysis research strategy is proposed.Through data mining strategies such as multivariate statistics,variable selection,enrichment analysis,database association analysis,variable selection and enrichment analysis,the association between different omics data(metabolomics and proteomics)are established for the metabolomics data of cerebral cortex and hippocampus of rats with traumatic brain injury model.This mining strategy not only saves a lot of time and research cost,but also expands and enriches the mining merit of omics data,and avoids only computer analysis results for the correlation between multi-source data.Based on this strategy,we screen the metabolites of different brain tissues(cortex and hippocampus)involved in the pathophysiological imbalance of traumatic brain injury,mine the proteins associated with metabolites in different brain tissues,interpret their biological functions,and predict the correlation role of multiple groups in the physiological and pathological mechanism of traumatic brain injury,providing more comprehensive and dynamic new data for the pathophysiological mechanism of traumatic brain injury.