Efficient gene set analysis of high-throughput data : From omics to pathway architecture of health and disease

Show full item record



Permalink

http://urn.fi/URN:ISBN:978-951-51-6288-5
Title: Efficient gene set analysis of high-throughput data : From omics to pathway architecture of health and disease
Author: Mishra, Pashupati
Contributor: University of Helsinki, Faculty of Biological and Environmental Sciences
Doctoral Programme in Integrative Life Science
Department of Clinical Chemistry, Fimlab Laboratories and Finnish Cardiovascular Research Center-Tampere, Faculty of Medicine and Health Technology, Tampere University
Publisher: Helsingin yliopisto
Date: 2020-08-06
URI: http://urn.fi/URN:ISBN:978-951-51-6288-5
http://hdl.handle.net/10138/316898
Thesis level: Doctoral dissertation (article-based)
Abstract: Background: A wide range of diseases, normal variations in physiology and development of different species are caused by alterations in gene regulation. The study of gene expression is thus crucial for understanding both normal physiology and disease mechanisms. High-throughput mea- surement technologies allow the profiling of tens of thousands of genes simultaneously. However, the high volume of data thus generated poses methodological challenges in inferring biological consequences from gene expression changes. Traditional gene wise analysis of high dimensional data is overwhelming, prone to noise and unintuitive. The analysis of sets of genes (gene set analysis, GSA), solves the problem by boosting statistical power and biological interpretability. Despite more than a decade of research on gene set analysis, there are still serious limitations in the existing methods. Aims of the study: The objectives of this study were: (1) development of an efficient p-value estimation method for GSA; (2) development of an advanced permutation method for GSA of multi-group gene expression data with fewer replicates; and (3) implementation of the developed methods for the identification of novel smoking induced epigenetic signatures at biological pathway level. Materials and methods: The first study involved the assessment of four different statistical null models for modeling the distribution of gene set scores calculated with the Gene Set Z-score (GSZ) function from permuted gene expression data. A new GSA method - modified GSZ (mGSZ) - based on GSZ and the most optimal distribution model was developed. mGSZ was evaluated by comparing its results with seven other popular GSA methods using four different publicly available gene expression datasets. The second study involved the evaluation of six different permutation schemes for GSA of multi-group (more than two groups) datasets based on the identification of reference gene sets generated using a novel data splitting approach. A new GSA method based on a modification of mGSZ (mGSZm) was developed by implementing the best permutation method for the analysis of multi-group data with fewer than six replicates per group. mGSZm was evaluated by contrasting its performance with seven other state-of-the-art GSA methods suitable for multi-group data. The evaluation was based on three different publicly available multi-group datasets. The third study involved an implementation of mGSZ for GSA of genome-wide DNA methylation data from the Cardiovascular Risk in Young Finns study (YFS) cohort with gene sets downloaded from the Molecular Signature Database (MSigDB). Methylation measurements were done on a subset of 192 individuals from whole-blood samples from the 2011 follow-up study using Illumina Infinium HumanMethylation450 BeadChips. Results: Overall, efficient and robust GSA methods were developed (studies I-II) and implemented (study III). In study I, the results demonstrated a clear advantage of asymptotic p-value estimation over empirical methods. mGSZ, a GSA method based on asymptotic p-values, requires fewer permutations which speeds up the analysis process. mGSZ outperformed state-of-the-art methods based on three different evaluations with three different datasets. In study II, results from a novel evaluation approach with two different datasets suggested that the proposed advanced permutation method outperformed the naive permutation method in GSA of multi-group data with fewer than six replicates. Evaluation of mGSZm, a GSA method equipped with the advanced permutation method and asymptoticn/a
Subject: genetics
Rights: This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.


Files in this item

Files Size Format View

There are no files associated with this item.

This item appears in the following Collection(s)

Show full item record