Section: Scientific Foundations

Comparative analysis and Noncoding RNAs

Noncoding RNA analysis is another fundamental topic for Sequoia . Our first publications on this subject date back to 2003. First, we proposed a new method for RNA structure inference and implemented it in a program called carnac . carnac has been evaluated as one of the best available software tools for comparative RNA structure prediction in an independent comparative survey [43] . It also gave rise to a thorough presentation in a recent survey paper [50] , and has been the subject of an invited book chapter in a general-purpose bioinformatics manual  [62] . Since its first release, the caRNAc software has been regularly updated. Second, we worked on theoretical models for RNA comparison. Comparison of RNA structures should take into account several levels of information corresponding to hierarchical RNA folding: sequence, secondary structure, tertiary interactions, involving complex combinatorial objects. RNA structures are usually modelled by ordered labeled trees or graphs, such as arc-annotated sequences. Our work led to substantial advances on tree edit distance algorithms [4] [63] , [54] , tree models [19] , [18] and comparison of arc-annotated sequences  [36] , [35] .

At the international level, a large number of groups develop similar researches on bioinformatics for noncoding RNAs, amongst them the Institut für theoretische Chemie, from Vienna University (I. Hofacker), the group of Practical Computer Science from Universität Bielefeld (R. Giegerich), the Bioinformatics Research Group from University of Copenhaguen (J. Gorodkin), the Computer Science Group from Ben Gurion University (M. Ziv-Ukelson). The specificity of Sequoia lies on a strong algorithmic background along with realistic biology models.