Development of novel tools for cancer diagnosis, prognosis and treatment using intra- or inter-species transcriptome metaanalysis
Author(s)
Advisor
Date
2017-09Publisher
Bilkent University
Language
English
Type
ThesisItem Usage Stats
151
views
views
36
downloads
downloads
Abstract
In the past decades, a considerable number of studies have performed meta-analysis on
large data collections to prioritize sets of genes, pathways or types/categories of disease
focusing on either differential expression, survival analysis, or co-expression networks.
However, not many web applications or databases have been developed from these
studies thus findings largely remained restricted to the addressed questions and it was
not possible for other researchers to use the collected data for the evaluation of novel
hypotheses. In this thesis, transcriptomic meta-analysis strategies have been applied to
untangle complexities in multiple aspects of cancer research including treatment,
diagnosis, and prognosis. Furthermore, three different web-tools have been developed
which are not limited to a single type of meta-analysis. In this context, in addition to
interesting cancer related findings, novel methodologies have been proposed and tested
in the field of meta-analysis and cancer research. First chapter of the thesis presented a general introduction on the concepts of the thesis.
Second chapter focused on a pathway comparison strategy based on meta-analysis that
was used to reveal concordant/discordant aspects of rapamycin-mediation on
transcriptomes of zebrafish and mouse. Analysis has shown that ribosomal terms were
significantly upregulated while proteasome was downregulated in both species. Zebrafish has undergone a whole-genome duplication event; I also found out that
rapamycin treatment resulted in largely concordant behavior of duplicated gene pairs. In
addition, an online database, CompariZome, was developed to evaluate the duplicate
zebrafish gene pairs in multiple GEO datasets in zebrafish in comparison to respective
human expression datasets. In the third chapter of this thesis, I focused on identification
of correlation between a trio of genes, CDH1, HNF4A, and GRHL3, using Cancer Cell
Line Encyclopedia (CCLE) dataset to reveal the significance of association between
these genes in different cancers, including breast and other epithelial cancers. The
findings indicated correlation within the module and has demonstrated the power of
meta-analysis using CCLE dataset. In the fourth chapter of this thesis, I focused on
understanding the association of CHRNA5, a subunit of cholinergic receptors, with
epithelial-to-mesenchymal transition (EMT) as well as epithelial differentiation, TP53 induction, and estrogen (E2) signaling with respect to breast cancer. Meta-analysis of invitro
and in-vivo microarray expression datasets showed that CHRNA5, itself, and its
positively co-expressed neighbors, were likely secondary targets of E2-signaling;
overexpressed in ER- breast cancer patients; and indicators of worse prognosis.
Functional annotation revealed that CHRNA5 and its co-expression network was indeed
associated with proliferation related pathways. Based on meta-analysis of different
cohorts processed in the study, an online database E2S (Estrogen (E2) to Survival) was
developed that can facilitate user to query any gene for evaluation of E2-mediated
effects, regulation by estrogen receptor (ER), prognostic importance and co-expression
network along with functional annotations. In the fifth chapter of this thesis I focused on deciphering the correlation and
deregulation between a human parolog pair of genes, i.e., mineralocorticoid and
glucocorticoid receptors (MR and GR, respectively) in breast cancer. Meta-analysis of a
separate normal/tumor cohort revealed that both genes were downregulated in breast
cancer and their expressions were highly positively correlated. However, deregulation
analysis predicted that expression of MR and GR was more tightly regulated in normal
breast hence its regulation might be lost with the onset of tumorigenesis. Another Shiny
database, DualExpBC, was developed to evaluate differential expression of a gene in
breast cancer as well as correlation and deregulation of expression between any two input genes in the breast normal/cancer expression cohort. With this thesis, I have
developed novel tools and approaches for intra- and inter-species comparative
transcriptomics and meta-analysis providing potential diagnostic, prognostic and
therapeutic biomarkers.
Keywords
Meta-analysisCo-expression network
Breast cancer
E2 signaling
CHRNA5
Inter-species and Intra-species transcriptome analysis
Mineralocorticoid and Glucocorticoid receptors