Section: New Results

RNA

To mitigate the current absence of a selective scientific event dedicated to RNA computational biology, impeding the dissemination of recent methodological results, Amib members have participated in the creation of the Computational Methods for Structural RNAs workshops (Cmsr'14 ). This first installment of the event was hosted in Strasbourg as a workshop of the 2014 edition of European Conference on Computational Biology. Its proceedings were published by McGill University [33] , and extended versions of selected articles were invited to appear in the Journal of Computational Biology.

RNA visualization

The field of RNA visualization is now rich with multiple tools that accommodate different needs, arising from a variety of application contexts. In order to help end-users navigate through the jungle of available options, Y. Ponty and F. Leclerc (Igm , Univ. Paris-Sud) have contributed a review of existing tools, and illustrate their usage to address a collection of typical use-cases [35] .

RNA design and structures

The past couple of years have seen the multiplication of heuristic or exponential time algorithms for the RNA design problem. This situation motivates a survey, which s currently lacking, that would focus on the relative merits of existing algorithms, and assess their applicability towards the typical goals of synthetic biology. Such an objective evaluation is at the core of the PhD project of Vincent Le Gallic, which was started in September 2014.

With Antoine Soulé, a PhD student of J-M Steyaert and J. Waldispühl (McGill), a comparative study of the various softwares for the inverse RNA folding problem is under revision and a new version of RNAmutant in the langage GAP-L with enrichment has been designed.

Besides, we have published a general survey on RNA structure comparison [9] .

RNA splicing regulation

RNA splicing is a modification of the nascent pre-messenger RNA (pre-mRNA) transcript in which introns are removed and exons are joined. The U2AF heterodimer protein has been well studied for its role in defining functional 3’ splice sites in pre-mRNA splicing, but multiple critical problems are still outstanding, including the functional impact of their cancer-associated mutations. In collaboration with Xiang-Dong Fu's groups in San Diego and Wuhan, , through genome-wide analysis of U2AF-RNA interactions, we reported in [16]  that U2AF has the capacity to define  88% of functional 3’ splice sites in the human genome. Numerous U2AF binding events also occur in other genomic locations, and metagene and minigene analysis suggests that upstream intronic binding events interfere with the immediate downstream 3’ splice site associated with either the alternative exon to cause exon skipping or competing constitutive exon to induce inclusion of the alternative exon.

RNA 3D structure modelling

Conformational diversity for RNA ensemble analyses is often provided by sophisticated molecular dynamics simulations. Long trajectories with specialized force fields on dedicated supercomputers are required to adequately sample conformational space, limiting ensemble analyses to modestly-sized RNA molecules. To avoid these limitations, we developed an efficient conformational sampling procedure, Kino-geometric sampling for RNA (KGSrna), which can report on ensembles of RNA molecular conformations orders of magnitude faster than MD simulations. In the KGSrna model, the RNA molecule is represented with rotatable, single bonds as degrees-of-freedom and groups of atoms as rigid bodies. In this representation, non-covalent bonds form distance constraints, which create nested, closed cycles in a rooted spanning tree. Torsional degrees-of-freedom in a closed ring demand carefully coordinated changes to avoid breaking the non-covalent bond, which greatly reduces the conformational flexibility. The reduced flexibility from a network of nested, closed rings consequently deforms the biomolecule along preferred directions on the conformational landscape. This new procedures projects degrees-of-freedom onto a lower-dimensional subspace of the conformation space, in which the geometries of the non-covalent bonds are maintained exactly under conformational perturbation. The dimensionality reduction additionally enables efficient exploration of conformational space and reduces the risk of overfitting sparse experimental data. Kinogeometric sampling of 3D RNA models can recover the conformational landscape encoded by proton chemical shifts in solution and is thus of great help to interpret NMR experimental data [11] . The computational efficiency of this approach, combined to its inherent parallel nature could also be adapted to model large assemblies on parallel platforms.

Our expertise was also essential in modelling junction of the RNA structure of a large biomolecule of interest, the tmRNA so as to study its interaction with the SmpB protein. Results obtained in collaboration with experiementalists, mainly P. Vachette at Ibbmc and S. Nonin-Lecomte at the Lcrb were made available in [15] .