Comparative analysis of methods for microbiome study

Iyer, Mihir Vishwanath

Permalink

https://hdl.handle.net/2142/108457 Copy

Description

Title

Comparative analysis of methods for microbiome study

Author(s)

Iyer, Mihir Vishwanath

Issue Date

2020-07-08

Director of Research (if dissertation) or Advisor (if thesis)

Iyer, Ravishankar K

Department of Study

Electrical & Computer Eng

Discipline

Electrical & Computer Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Date of Ingest

2020-10-07T20:59:39Z

Keyword(s)

• microbiome
• data science

Abstract

Microbiome analysis is garnering much interest with benefits including improved treatment options, enhanced capabilities for personalized medicine, greater understanding of the human body, and contributions to ecological study. Data from these communities of bacteria, viruses, and fungi are feature rich, sparse, and have sample sizes not appreciably larger than the feature space, making analysis challenging and necessitating a coordinated approach utilizing multiple techniques alongside domain expertise. This thesis provides an overview and comparative analysis of these methods, with a case study on cirrhosis and hepatic encephalopathy demonstrating a selection of methods. Approaches are considered in a medically motivated context where relationships between microbes in the human body and diseases or conditions are of primary interest, with additional objectives being the identification of how microbes influence each other and how these influences relate to the diseases and conditions being studied. These analysis methods are partitioned into three categories: univariate statistical methods, classifier-based methods, and joint analysis methods. Univariate statistical methods provide results corresponding to how much a single variable or feature differs between groups in the data. Classifier-based approaches can be generalized as those where a classification model with microbe abundance as inputs and disease states as outputs is used, resulting in a predictive model which is then analyzed to learn about the data. The joint analysis category corresponds to techniques which specifically target relationships between microbes and compare those relationships among subpopulations within the data. Despite significant differences between these categories and the individual methods, each has strengths and weaknesses and plays an important role in microbiome analysis.

Graduation Semester

2020-08

Type of Resource

Thesis

Permalink

http://hdl.handle.net/2142/108457

Copyright and License Information

Copyright 2020 Mihir Iyer

Owning Collections