RNA editing identification in transcriptomics data and assessment of impact in miRNA targeting

Postgraduate Thesis uoadl:2895092 254 Read counter

Unit:
Κατεύθυνση Βιοπληροφορική
Πληροφορική
Deposit date:
2020-01-28
Year:
2020
Author:
Miliotis Marios
Supervisors info:
Καθ. Άρτεμις Χατζηγεωργίου, Καθηγήτρια Βιοπληροφορικής, Τμήμα Μηχανικών Η/Υ, Τηλεπικοινωνιών και Δικτύων του Πανεπιστημίου Θεσσαλίας
Δρ. Martin Reczko, Ερευνητής Καθηγητής, Ερευνητικό Κέντρο Βιοϊατρικών Επιστημών “Αλέξανδρος Φλέμινγκ” Δρ. Αλέξανδρος Δημόπουλος, Μεταδιδακτορικός ερευνητής, Ερευνητικό Κέντρο Βιοϊατρικών Επιστημών “Αλέξανδρος Φλέμινγκ”
Original Title:
Εντοπισμός RNA τροποποιήσεων σε δεδομένα μεταγραφώματος και εκτίμηση της επίδρασής τους στους στόχους των miRNA
Languages:
Greek
Translated title:
RNA editing identification in transcriptomics data and assessment of impact in miRNA targeting
Summary:
RNA editing is a co/post-transcriptional process, during which an RNA molecule is undergone an alteration of its sequence by insertion, deletion or modification. The majority of such changes in metazoans is comprised by adenosine (A) to inosine (I) nucleotide transitions, which are catalyzed by members of the adenosine deaminase gene family (ADAR) acting on double-stranded RNA (dsRNA). RNA editing is relatively widespread in Alu-containing mRNA molecules. Editing of the coding sequence in pre-mRNAs can modify codons and lead to the incorporation of different amino acids during translation, contributing to protein function diversity. However, most A-to-I editing events occur in non-coding regions of pre-mRNAs and mRNAs, as well as in non-coding RNAs. Editing in the UTR (untranslated region) of mRNAs can regulate their translation, splicing and degradation. Also, events in microRNA (miRNA) and long non-coding RNA (lncRNA) sequences, as well as their binding sites, can affect their biogenesis, target recognition, structure and stability. The goal of this study was to compare a set of RNA editing identification tools, distinguish true substitution events in 3’UTR of mRNAs and assess their impact on miRNA specificity and binding efficacy. Initially, we used matching RNA and DNA sequencing data to identify A-to-I RNA editing events in 3’UTR regions. This was done to investigate event calls in individual level, increasing specificity. Our dataset consisted of 2 test samples, 1 ADAR enzyme knockdown control sample and RADAR, a comprehensive collection of A-to-I editing events in human, mouse and fly transcripts, with the last two resources being used as ground truth. Then we went on to find the best algorithm to identify events. The ADAR knockdown dataset was useful to pinpoint high false positive rates. The comparison included RES-Scanner employing the Burrows-Wheeler Aligner (BWA), REDItools running with GSNAP aligner and RNAEditor which was run with BWA (default option) and GSNAP. The first two approaches natively support paired/matched RNA-DNA datasets, while the latter was modified to include DNA information. The most robust behaviour in terms of sensitivity and specificity was observed from RNAEditor with BWA aligner. Edited and non-edited 3’UTR were subsequently used as input in miRNA target prediction algorithms to statistically assess differences in the computed binding sites that arose due to the editing phenomena. The wildtype counterpart of the ADAR knockdown experiment was employed here to further enhance the comparison. Events annotated in 3’UTR were used to generate 2 equally numbered sets of sequences, 2062 of which belonged to highly repetitive Alu regions and 144 in non-Alu. The top 50 expressed miRNA in each sample were used to confine the target prediction analysis that was performed using TargetScan and MIRZA-G algorithms. Both of them were run without incorporating evolutionary features, which cannot be effectively measured in the case of the edited sequences. The results show a strong preference towards modification of binding site feature distributions, rather than generating new or depleting existing sites. Moreover, we observed a mild alteration of the repressive action of miRNA targeting edited UTR. A separate analysis of highly edited UTR, i.e. UTR subjected to multiple editing events, did not indicate any correlation with the degree of the change. The lack of a global trend in the alteration of miRNA repressive activity implies RNA editing can serve distinct roles in miRNA efficacy, fine-tuning their targeting action on a case-by-case basis.
In this study, we did a benchmark of RNA editing identification tools, we came up with a set of edited UTRs and performed miRNA target prediction on them. This analysis indicated alteration of the targeting efficacy by miRNA, irrespective of the number of editing events in the region. Further analyses of more samples and conditions will be useful to validate and empower our findings.
Main subject category:
Science
Keywords:
RNA editing, NGS, transcriptomics, A-to-I substitution, miRNA target prediction
Index:
Yes
Number of index pages:
9
Contains images:
Yes
Number of references:
98
Number of pages:
109
Miliotis_ITMB_thesis.final.pdf (4 MB) Open in new window