Skip to content


Evolution and Conservation of RNA structures

RNA viruses exhibit high mutation rates, rapid evolution, and frequent host switching, which make them particularly challenging to control. By studying the evolution of RNA structures in viruses, we can identify molecular mechanisms that contribute to viral adaptation and emergence, and develop strategies to prevent or treat viral infections.

We investigate the evolutionary relationships between RNA viruses by analyzing conserved RNA structures in their genomes. RNA structures play critical roles in viral replication, gene expression, and pathogenesis, and can evolve rapidly in response to environmental pressures. By analyzing conserved RNA structures in multiple sequence alignments of viral genera or families, we can gain insights into the functional constraints acting on different regions of the viral genome, and identify potential targets for antiviral therapies. We further use computational methods to predict RNA secondary structures de novo, which can reveal novel structural elements that are not yet described. By comparing the predicted structures to those of phylogenetically viruses, we can identify conserved and divergent structural elements, and gain insights into their functional roles.

RNA-Protein Interaction predictions

RNA-protein interactions play a crucial role in many biological processes, including gene expression, RNA splicing, RNA localization, and protein synthesis. These interactions are essential for proper cellular function and dysregulation of RPIs can lead to a range of diseases, including cancer and neurodegenerative disorders. RNA-protein interactions also play a critical role in host-pathogen interactions, where viral and bacterial proteins interact with host RNA molecules to hijack cellular machinery and evade host immune responses. Understanding these interactions is crucial for developing effective treatments and vaccines for infectious diseases.

One of the main challenges in RPI prediction is the limited availability of experimental data on RPIs, especially for non-model organisms. Most existing RPI prediction tools rely on machine learning algorithms that are trained on a set of known RPIs, typically obtained from model organisms or well-studied systems. These tools can be highly accurate when applied to similar systems, but they may not generalize well to non-model organisms or new types of RPIs.

To address these challenges, researchers are developing new approaches for RPI prediction that leverage large-scale data resources, such as transcriptome and proteome data, to identify conserved features that are predictive of RPIs. We are developing a workflow to determine such conserved features using evolutionary information of homologous RNA and protein sequences, respectively. In their respective MSAs, we look for signals of co-evolution and correlate the present mutations with binding affinities and preferences known from the literature or derived from experiments conducted by our collaborators.

Incongruent Evolution

When analyzing RNA consensus structures, sometimes closely related viruses exhibit no conserved structure at all. However, folding each sequence individually exposes the exact same structural conformation, but shifted within the sequence. This phenomenon is called incongruent evolution, and essentially describes the effect of homologous nucleotides not contributing to homologues base pairs.

There is some speculation that incongruent evolution may provide a selective advantage for pathogens by allowing them to evolve more complex RNA structures that are better able to evade host immune responses or manipulate host cellular machinery. Incongruent evolution may also be advantageous for pathogens that rely on RNA-RNA interactions, such as riboswitches or RNA-based regulatory elements, as it may allow these elements to evolve more rapidly and acquire novel functions.

Recently, a model and an algorithm to detect incongruent evolution between two RNA sequences have been proposed. Together with collaborators, we apply and extend this model to pathogens and systematically scan their genomes for regions with incongruent evolution to further our understanding of this peculiar mechanism.

Beyond the canonical

Pseudoknots are secondary structural motifs that arise when two distinct regions of an RNA molecule fold back on each other and form nested base pairs, creating a knot-like structure. Pseudoknots are often found in the non-coding regions of RNA molecules, such as the 5' and 3' UTRs, and can play a variety of roles in RNA function, including regulating translation and RNA stability, as well as facilitating RNA-protein interactions.

Flexible regions in RNA are regions of the RNA sequence that can adopt multiple structures with different functions. These regions can be highly dynamic and can switch between different structures depending on environmental cues or cellular signals. One example of a flexible region is the 3' X-tail in HCV, either serving for translation or replication of the virus genome.

Conventional algorithms and tools usually not incorporate such non-canonical elements or only specific cases of these (e.g., H-type pseudoknots). We are studying different approaches, such as ambivalent covariance models or principles from graph theory, to incorporate this additional layer of complexity and information into current models.

Beyond viruses

While our focus is on viruses, we are generally interested in pathogens and their mechanisms during infection. Thus, we also try to consider bacteria and fungi as well. So, if you are in general interested in our projects and approaches, but do not like viruses for some reason, do not hesitate to contact us.