Research For The Analysis Method Of The Disease Feature Of Omics Datasets

Over the last decades, the development of high-throughput techniques has a great influence on the disease research:In the study of human disease, more and more analysis methods for omics data are being applied and developed. These analysis methods improve the disease feature research into molecule level. Facing different kinds of omics data, how to mine the important information relative to disease has been a hot topic in bioinformatics. Mining the disease patterns in omics data is helpful to understand the mechanism of disease:On one hand, it can improve medical diagnosis and prediction accuracy. On the other, it will provide new ideas and methods for the intervention and treatment of diseases.In consideration of medical diagnosis, disease development prediction and disease feature pattern search, the dissertation carried out a few disease researches of analysis methods for omics data. These works’ contents include:(1) A mid-level fusion method for omics datasets was proposed to solve the classification problem by using many different kinds of omics datasets. To test the classification effectiveness, the method was compared with other fusion methods on two public cancer omics datasets.(2) A disease analysis prediction method based on hierarchical decision was provided to solve the disease development prediction problem by using different kinds of omics datasets. The analysis method was applied to predict the development of pre-diabetes patients and to build an omics feature correlation network. The network can provide a reasonable explanation from the view of biology.(3) To search for the disease patterns of omics features, disease omics feature pattern search methods were provided. These methods were based on the order of omics features. It was used on the omics data of some pre-diabetes and diabetes patients. The features found by the methods were connected with the development of pre-diabetes patients. From the biological point of view, the connection between metabolites’ concentrations and diabetes can interpret these results.