PhD Defence: Corinna Theis

2015-12-07: The potential of RNA 3D Modules in the prediction of structured RNAs. The defence will take place on Decemebr 7th at 13:00 in Orangeriet, Dyrlægevej 36, 1870 Frederiksberg C.

Everybody is welcome. Registration is not necessary. After the defence a reception will be held from 16.30-18:30 at Grønnegårdsvej 3, library on 2nd floor.

Assessment committee:

  • Associate Professor Jakob Hull Havgaard, Department of Veterinary Clinical and Animal Sciences, University of Copenhagen (Chairman)
  • Professor Janusz M. Bujnicki, Laboratory of Bioinformatics, Faculty of Biology, Adam Mickiewicz University, Poland
  • Professor Manja Marz, Faculty of Mathematics and Computer Science, Friedrich-Schiller-University Jena, Germany

Supervisor: Professor Jan Gorodkin, Center for non-coding RNA in Technology and Health, IKVH, KU.


Recent progress in predicting RNA structure is taking a route towards filling the 'gap'in two-dimensional (2D) RNA structure prediction where, for example, predicted internal loops often form non-canonical base pairs. This directs the focus on RNA three-dimensional(3D) modules. RNA 3D modules are small sets of non-canonical base pairs which form distinct patterns often located in internal or hairpin loops. They are important for the functionality of a molecule, e.g. they serve as local structural organizers as well as protein and RNA binding sites. A combination of secondary structure and 3D module predictions can approximate the spatial structure and thus indicate the functional mechanisms of RNAs.With the increasing amount of known 3D modules the general interest in matching them from one molecule to other molecules for which the 3D structure is not known increases.However, a major challenge is to determine whether the module is trustworthy in the first place. Another challenge is that module recognition and modeling requires time consuming manual interference.In a first approach we developed a pipeline, metaRNAmodules, to create probabilistic models of known and putative RNA 3D modules. The pipeline completely automates extracting characteristic features of 3D modules from solved tertiary structures and mapping of such modules to alignments to obtain comparative evidence across multiple organisms. Some of the gained 3D modules are already established, but there might be prospective sites which have not be identified yet to be functional. We show that the automatically generated models have discriminative power and are able to find already known 3D modules. The models can be used to scan single sequences or multiple sequence alignments for further occurrences of a particular module. The source code of the pipeline is available at Further, the new models are used to scan two different data sets on a genome-wide scale. First, the models are applied on a sequence-based whole genome alignment of vertebrates in conjunction with a screen for structured RNAs including ncRNAs and cis-regulatory elements in mRNAs. This is a proof of concept that the combination of RNA 3D modules and RNA secondary structure predictions help to improve the prediction accuracy and lower the false discovery rate of such screens. Secondly, structure-based vertebrate alignments are screened to predict further occurrences of 3D modules. The complementarity of the 3D modules and the secondary structure predictions are quantified by a newly introduced measurement. This can aid in the identification of promising candidates, e.g. for follow-up experiments such as structure probing.