Date on Master's Thesis/Doctoral Dissertation

5-2025

Document Type

Doctoral Dissertation

Degree Name

Ph. D.

Department

Interdisciplinary Studies

Degree Program

Interdisciplinary Studies with a specialization in Bioinformatics, PhD

Committee Chair

Barve, Shirish

Committee Member

Rouchka, Eric

Committee Member

Ghare, Smita

Committee Member

Smith, Melissa

Committee Member

McClain, Craig

Author's Keywords

gut microbiome; integrated data analysis; multi-omics; HIV-1 infection, alcohol use disorder, aging

Abstract

Gut dysbiosis characterized by reduced abundance of beneficial butyrate-producing bacteria has been independently linked to HIV-1 infection and heavy alcohol drinking. Further, gut dysbiosis results in loss of gut barrier integrity, microbial translocation and host-specific systemic inflammation. Therefore, to evaluate the functional consequences of structural changes in the gut microbiome, integrated data analysis is imperative. In this dissertation, we perform integrated analyses using data from multi-omics platforms to examine the structural and functional features of the gut microbiome of PWH. Gut microbiome composition was evaluated by sequencing V4 region of 16S rDNA, concentrations of metabolites and host-specific immune markers were measured by metabolomics and immune assays. To perform integrated analysis, we explore the use of Data Integration Analysis of Biomarker discovery using Latent cOmponents (DIABLO; mixOmics) and Weighted Gene Co-expression Network Analysis (WGCNA). Additionally, we also present an automated pipeline called ‘buty_cat’ that integrates information from 16S rDNA sequencing with inferred metagenomics analysis using PICRUSt2. Employing a comprehensive butyrate catalogue published in literature, ‘buty_cat’ can identify specific butyrate-producing genera that harbor genes involved in butyrate synthesis. Overall, our observations indicate that the ultimate choice of method to perform integrated analysis depends on the research question. Moreover, data pre-processing such as log transformation, data normalization, outlier detection as well as feature selection are crucial for the most optimum performance of any integrated data analysis model. Most importantly, integrated analysis provides a holistic overview of the clinical underpinnings of disease pathology that will help clinicians and researchers develop therapeutic treatment strategies.

Download

Share

COinS