Section: New Results

Sequence annotation

Participants : François Coste [contact] , Catherine Belleannée, Gaëlle Garet, Clovis Galiez, Laurent Miclet, Jacques Nicolas.

• Expressive pattern matching [C. Belleannée, J. Nicolas] We have presented for the first time Logol, a new application designed to achieve pattern matching in possibly large sequences with realistic biological motifs. Logol consists in both a language for describing patterns, and the associated parser for effectively scanning sequences (RNA, DNA or protein) with such motifs. The language, based on an high level grammatical formalism, allows to express flexible patterns (with misparings - improper alignment of DNA strands - and indels) composed of both sequential and structural elements (such as repeats or pseudoknots)[21] [Online publication] . Logol has been applied to the detection of -1 frameshifts, a structure including pseudoknots, on a reference benchmark (Recode2) [26] [Online publication] .

• Analysis of sequence repeats [J. Nicolas] We have participated to a book that introduces up-to-date methods for the identification and study of transposable elements in genomes. J. Nicolas contributed with a chapter that provides an overview of the formal underpinnings of the search for these highly repeated elements in genomic sequences and describes a selection of practical tools for their analysis. It concludes with the interest of syntactic analysis in this domain [24] [Online publication] .

• Grammatical models for local patterns [G. Garet, J. Nicolas, F. Coste] We studied the annotation of new proteins with respect to banks of already annotated protein sequences. For this task, we are developping grammatical inference methods. We introduced new classes of substitutable languages and new generalization criterion based on local substitutability concept and illustrated the great potential of the approach on a benchmark considering a real non trivial protein family. [16] [Online publication]

• Local maximality [L. Miclet] Starting from locally maximal subwords and locally minimal superwords common to a finite set of words, we have defined the corresponding sets of alignments. We gave a partial order relation between such sets of alignments, as well as two operations between them and showed it has a lattice structure that can be used for inducing a generalization of the set of words [18] [17] .

• Searching for Smallest Grammars on Large Sequences and Application to DNA [F. Coste] We are motivated by the inference of the structure of genomic sequences, that we address as an instance of the smallest grammar problem. Previously, we reduce it to two independent optimization problems: choosing which words will be constituents of the final grammar and finding a minimal parsing with these constituents. This year we made these ideas applicable on large sequences. First, we improved the complexity of existing algorithms by using the concept of maximal repeats for constituents. Then, we improved the size of the grammars by cautiously adding a minimal parsing optimization step. Together, these approaches enabled us to propose new practical algorithms that return smaller grammars (up to 10%) in approximately the same amount of time than their competitors on a classical set of genomic sequences and on whole genomes. [14] [Online publication] .

• CyanoLyase: a database of phycobilin lyase sequences, motifs and functions [F. Coste] In collaboration with our partners of the ANR project Pelican, we have set up CyanoLyase (http://cyanolyase.genouest.org/ ), a manually curated sequence and signature database of phycobilin lyases and related proteins. Protomata-Learner has been used to establish the signature of the 32 known subfamilies that are used to rapidly retrieve and annotate lyases from any new genome [13] [Online publication]