[Internship/PhD] Algorithmic aspects of RNA locally optimal structures

Scientific context. RiboNucleic Acids (RNAs) are single-stranded macromolecules which can be crudely described as sequences of length ranging from 20 to 3 000 letters, over the four-letters alphabet {A,C,G,U}. Due to its single-stranded nature, any synthesized RNA molecule undergoes a folding process, in which connections, also known as base-pairs mediated by hydrogen bonds, are established between its nucleotides. The outcome of RNA folding is a large variety of structures, or spatial conformation, which strongly determine the function of a given RNA within cellular mechanisms. From a computer science perspective, the structure(s) of an RNA can be acceptably abstracted as a graph, or a tree.

Internship objective. The main goal of this internship is the design and implementation of one (or several) algorithm(s) that would enable the enumeration and random generation of RNA locally optimal structures. Such structures act as kinetic traps in the energy landscape, aka local minima of the energy function, and are generally believed to significantly slow-down, or even disrupt, the folding of structured RNAs. Their enumeration represents a mandatory first step towards an efficient in silico analysis of the kinetic behavior of RNA from its sequence. Previous works [1] have resulted in a polynomial-time algorithm for the enumeration of locally optimal structure in a combinatorial setting based on base-pair maximization. This algorithm will be the starting point of the internship, and it will be completed/extended to capture the complete set of features supported by the – more accurate and realistic – Turner model [2].

Environment. The intern will be hosted by the Bonsai team (Inria Lille Nord Europe et LIFL), and co-supervised by Hélène Touzet and Yann Ponty (CNRS LIX/PIMS Vancouver). It may start between January and April 2015, and will last from 4 to 6 months. A compensation (gratification de stage mensuelle) may be allocated to the successful candidate.
This work is part of the French/Austrian project RNALands, recently funded by ANR and FWF, whose aim is the design of efficient predictive methods for RNA kinetics. It may be followed by a PhD at LIX (Computer Science Dept of Ecole Polytechnique –Palaiseau), and include research visits to the other partners of the project, in Lille and Vienna.

Candidate profile. The perfect candidate for this internship is a fifth-year student in Computer Science with a strong background in algorithms/data structures and/or bioinformatics, and a real taste for algorithm design and implementation. A preliminary experience in C/C++, plus a scripting language of the candidate’s choice (Python, Ruby, Perl…), is required. A background in Molecular Biology/Biochemistry, or some measurable level of intellectual curiosity for the subject, will be considered a plus.
To apply, please send:

  • a complete resume;
  • a cover letter stating your objectives and expectations;
  • a copy of your academic record/transcript for the past two years;
  • the contact of 2-3 reference;

to Hélène Touzet (helene.touzet@lifl.fr) and Yann Ponty (yann.ponty@lix.polytechnique.fr).


  • [1] Azadeh Saffarian, Mathieu Giraud, Antoine de Monte, and Hélène Touzet. RNA locally optimal secondary structures. J Comput Biol, 19(10):1120–1133, Oct 2012.
  • [2] D. H. Mathews, J. Sabina, M. Zuker, and D. H. Turner. Expanded sequence dependence of thermodynamic parameters improves prediction of RNA secondary structure. J Mol Biol, 288(5):911–940, May 1999.

Leave a Reply

Your email address will not be published.