Back to search

University of Thessaly (UTH)

Αλγόριθμοι για την υπολογιστική ανάλυση της λειτουργίας των μη κωδικών μεταγράφων

Abstract

dc:description

The RNA revolution has turned non-coding RNA (ncRNA) from dark-matter into a biological research hotspot. Accumulating evidence from multiple Next Generation Sequencing (NGS) experiments has recently introduced the regulatory roles of ncRNAs in a wide range of biological processes. This thesis focuses on the development of computational algorithms for the functional characterization of non-coding transcripts, while investigating in-depth their in-between interactions. The methodologies developed during this thesis combine advanced next-generation sequencing (NGS) data analyses and state-of-the-art Machine Learning algorithms in order to perform automated analyses and to monitor the corresponding results.This doctoral thesis studies specific categories of RNA transcripts: microRNAs (miRNAs) and long non-coding RNAs (lncRNAs). miRNAs are single stranded RNA molecules approximately 22 nucleotides long. They have been deemed central post-transcriptional gene regulators and play a key role in numerous biological processes. Therefore, miRNAs are intensively studied for their potential as biomarkers and/or therapeutic targets. Apart from their involvement in physiological processes, microRNAs appear to be associated with a plethora of pathological conditions. Although microRNAs are mainly considered mRNA repressors, there are studies supporting miRNA-lncRNA interactions. lncRNAs are long non-coding transcripts that can also regulate gene expression. To this end, DIANA-LncBase database was designed in order to characterize the entire spectrum of miRNA interactions with lncRNAs. LncBase supports a compendium of experimentally supported miRNA-lncRNA interactions. It contains more than 70,000 interactions derived from the analysis of numerous NGS experiments and specific low-throughput techniques, across 66 different types spanning 36 tissues in human and mouse species. DIANA-TarBase update was also part of the thesis. TarBase v7 is considered the largest available repository of miRNA-mRNA interactions as compared to any of the relevant databases. It hosts more than half a million interactions from published experiments on 356 different cell types (59 tissues), belonging to 24 species. The detailed cataloguing of RNA interactions unveiled a set of approximately 400 unique viral-miRNA:lncRNA interacting pairs in human virus-infected cells. This type of regulation adds an extra layer of complexity in the miRNA interactome, and perplexes the network with the inclusion of virus-encoded and human transcript interactions.By analyzing more than 150 raw CLIP-Seq datasets, DIANA-TarBase v7.0 and DIANA-LncBase are the first relevant databases to provide an unprecedented amount of experimentally supported interactions in many different cell types and tissues. Furthermore, RNA sequencing data were analyzed to accurately assess miRNA and transcript expression in the investigated cell types. Optimized pipelines were developed for the analysis of sequencing data, while a machine learning approach has been applied for the identification of miRNA binding sites. The adopted methodology for AGO-CLIP-Seq data analysis was compared against other available state-of-the-art implementations and has been proven robust and advantageous. During the course of the Doctoral thesis, the continuous archiving of experimental data from low and high-throughput methodologies, along with extensive evaluation of the available AGO-CLIP-Seq analysis programs, revealed that there was room for algorithms’ further improvement and optimization. State-of-the-art CLIP-guided target identification implementations currently manage to identify approximately half of the experimentally validated binding sites. To this end, a novel algorithm was developed for CLIP-Seq data analysis. The algorithm was trained and extensively tested on a comprehensive collection of accurate positive and negative miRNA-target interactions from numerous experimental data sources. It was additionally evaluated against all leading implementations, including CLIP-Seq analysis adopted by TarBase/LncBase. The results depict that the new approach not only significantly outperforms other implementations in terms of accuracy but also manages to increase sensitivity, predicting sites that were not detected by any other algorithm.The functional significance of miRNA interactions with coding and non-coding transcripts was further assessed with the evolutionary conservation of the miRNA binding sites. The thesis additionally associates the catalogued interactions to diseases and molecular pathways, providing new insights in ncRNA function. DIANA-microT web server was upgraded and enhanced with automated analyses pipelines (workflows) that can be applied to NGS-derived data. The ready-to-use modules seamlessly integrate DIANA supported algorithms for the identification of miRNA-gene interactions and miRNA-targeted pathway analyses.During the course of the Doctoral thesis, the candidate took part in 8 scientific studies involving computational approaches for determining the activity of the non-coding transcripts and in four of them the candidate is first author. The studies are published in international peer-reviewed scientific journals, while the total citations received to date are 310.

Degree

thesis:*
Grantor dc:publisher
University of Thessaly (UTH)
Year dc:date
2016

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Paraskevopoulou, Maria
  • Παρασκευοπούλου, Μαρία

Subjects

dc:subject × 10

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/38945

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

Paraskevopoulou, Maria; Παρασκευοπούλου, Μαρία. Αλγόριθμοι για την υπολογιστική ανάλυση της λειτουργίας των μη κωδικών μεταγράφων. University of Thessaly (UTH), 2016. http://hdl.handle.net/10442/hedi/38945