Back to search

University of Thessaly (UTH)

Μεθοδολογίες υπολογιστικής ανάλυσης και μετα-ανάλυσης δεδομένων γονιδιακής έκφρασης

Abstract

dc:description

This PhD thesis dealt with meta-analysis methodologies and its statistical techniques and applications in high-throughput experiments, and in particular on gene expression data from microarray experiments. Meta-analysis is a valuable tool for the synthesis of evidence across a wide range of study types including high-throughput experiments such as genome-wide association studies (GWAS) and gene expression studies. One of the issues that this thesis dealt with, was the investigation of optimal meta-analysis techniques. Various tests were performed on a relatively small and a larger random testing dataset (99 and 462 individuals respectively) using the old but classic method of Hedges correction and investigating whether better results can be obtained using the resampling method Bootstrap (of 200 or 500 repetitions). The results showed that the use of the Hedges correction is almost always necessary in small datasets. It was also investigated whether Bootstrap (200 repetitions) correction is giving better results, as being told, than using the Hedges correction only. Which was confirmed in the present study, however, it takes much longer. Point was also made that using 200 Bootstrap repetitions gives almost the same results in statistical power and utility with the time consuming 500 Bootstrap repetitions (which is considered more reliable in the literature). In the context of investigating various multifactorial diseases meta-analysis of gene expression data from microarray experiments related to preeclampsia was performed; to identify differentially expressed genes, to help elucidate the identification of the disease etiological mechanisms and even help deciphering differentiating mechanisms underlying the different subtypes of preeclampsia. According to this the databases Pubmed and GEO were searched for gene expression data on placental samples from case-control studies. Meta-analysis was performed, and 629 statistically significant differentially expressed genes were identified to be associated with preeclampsia at an FDR 1%. The identified differentially expressed genes were further analyzed with bioinformatics tools and we retrieve interaction networks and overrepresented biochemical pathways, in which they are taking part in. Network analysis showed few, highly interconnected, genes involved in innate immunity and signal transduction pathways which is indicative of a multifaceted disease with etiological heterogeneity. Over-representation analysis revealed that these genes participate mainly in carbohydrates, amino acids and pyrimidine metabolism, circadian clock system and signal transduction pathways. As noted, meta-analysis is a powerful tool for the synthesis of research studies. There are situations though, in which we have multiple outcomes or multiple treatments, in which multivariate meta-analysis framework which performs a joint modeling of the different quantities of interest could offer important advantages, such as increasing statistical power and allowing performing global tests. For these reasons, we adapted the multivariate meta-analysis methodology so that it could be applied in gene expression data, and we create an easy and reliable software, using the statistical package Stata. The implementation of the method is available at www.compgen.org/tools/multivariate-microarrays. Last but not least using the aforementioned software we performed a multivariate meta-analysis of microarray data in Crohn’s disease and ulcerative colitis, which are the main forms of Inflammatory Bowel Disease. They share similar symptoms but differ in the location and extent of inflammation and their complications. 249 differentially expressed genes were identified to be associated with Crohn’s disease and 38 with ulcerative colitis at an FDR of 1%. 20 of the differentially expressed genes were common to both diseases. A global test identified 260 differentially expressed genes associated with inflammatory bowel disease, 53 of which were not found in any of the disorders. With the use of bioinformatics methods important molecular pathways were identified, which could be implicated in the pathogenesis of inflammatory bowel disease, such as the JAK/STAT and interferon-gamma signaling pathways. Additionally genes involved in cell adhesion, apoptosis and carcinogenesis were found to be statistically important. Among others BCAT1 and GZMB are interesting novel differentially expressed genes that deserve further investigation in experimental models.

Degree

thesis:*
Grantor dc:publisher
University of Thessaly (UTH)
Year dc:date
2020

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Vennou, Konstantina
  • Βέννου, Κωνσταντίνα

Subjects

dc:subject × 25

Rights

dc:rights
Statement dc:rights
  • BY_NC_ND
Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/47552

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Vennou, Konstantina; Βέννου, Κωνσταντίνα. Μεθοδολογίες υπολογιστικής ανάλυσης και μετα-ανάλυσης δεδομένων γονιδιακής έκφρασης. University of Thessaly (UTH), 2020. http://hdl.handle.net/10442/hedi/47552