Integrative Statistical Learning with Applications to Predicting Features of Diseases and Health.

Integrative Statistical Learning in High-dimensional Time-series Data;Host Transcriptional Responses to Respiratory Viral Pathogens;Role of Hoxa9 in Leukemic Transformation;Spectral Analysis of Temporal Pathway Activity Using Graph Lapalacian;Information Geometric Analysis of Motif Profiles in ChIP-sequencing;Predictive Modeling and Classification in High-dimensional and Temporal Data;Biomedical Engineering;Genetics;Microbiology and Immunology;Pathology;Science (General);Statistics and Numeric Data;Health Sciences;Science;Bioinformatics

Huang, YongshengShedden, Kerby A. ;

University of Michigan

关键词: Integrative Statistical Learning in High-dimensional Time-series Data;    Host Transcriptional Responses to Respiratory Viral Pathogens;    Role of Hoxa9 in Leukemic Transformation;    Spectral Analysis of Temporal Pathway Activity Using Graph Lapalacian;    Information Geometric Analysis of Motif Profiles in ChIP-sequencing;    Predictive Modeling and Classification in High-dimensional and Temporal Data;    Biomedical Engineering;    Genetics;    Microbiology and Immunology;    Pathology;    Science (General);    Statistics and Numeric Data;    Health Sciences;    Science;    Bioinformatics;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/84435/huangys_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

This dissertation develops methods of integrative statistical learning to studies of two humandiseases - respiratory infectious diseases and leukemia. It concerns integrating statisticallyprincipled approaches to connect data with knowledge for improved understandingof diseases. A wide spectrum of temporal and high-dimensional biological and medicaldatasets were considered.The first question studied in this thesis examined host responses to viral insult. Ina human challenge study, eight transcriptional response patterns were identified in hostsexperimentally challenged with influenza H3N2/Wisconsin viruses. These patterns arehighly correlated with and predictive of symptoms. A non-passive asymptomatic state wasrevealed and associated with subclinical infections. The findings were validated and extended to three additional viral pathogens (influenza H1N1, Rhinovirus, and RSV). Theirdifferences and similarities were compared and contrasted. Statistical models were developedfor exposure detection and risk stratification. Experimental validations have been performed by collaborators at the Duke University.The second question studied in this thesis investigated the regulatory roles of Hoxa9and Meis1 in hematopoiesis and leukemia. Methods were developed to characterize theirglobal in vivo binding patterns and to identify their functional cofactors and collaborators.The combinatorial effects of these factors were modeled and related to specific epigeneticsignatures. A new biological model was proposed to explain their synergistic functions inleukemic transformation. Experimental validations have been performed by members of the Hess laboratory.Motivated by problems encountered in these studies, two algorithms were developedto identify spatial and temporal patterns from high-throughput data. The first method determinestemporal relationships between gene pathways during disease progression. It performs spectral analysis on graph Laplacian-embedded significance measures of pathway activity. The second algorithm proposes probabilistic modeling of protein binding events. Based on information geometry theory, it applies hypothesis testing coupled with jackknife-bias correction to characterize protein-protein interactions. Experimental validations were shown for both algorithms.In conclusion, this dissertation addressed issues in the design of statistical methodsto identify characteristic and predictive features of human diseases. It demonstrated theeffectiveness of integrating simple techniques in bioinformatics analysis. Several bioinformaticstools were developed to facilitate the analysis of high-dimensional time-series datasets.