Документ взят из кэша поисковой машины. Адрес оригинального документа : http://mccmb.belozersky.msu.ru/2013/abstracts/abstracts/154.pdf Дата изменения: Sun May 12 17:43:24 2013 Дата индексирования: Thu Feb 27 21:01:31 2014 Кодировка:

RNASurface: fast and accurate identification of motifs with high structural potential Ruslan Soldatov1, Svetlana Vinogradova1,2, Andrey Mironov1
1 2

,2

Institute for Information Transmission Problem, Moscow, Russia

Department of Bioengineering and Bioinformatics, Moscow State University, Moscow, Russia solrust@mail.ru

RNA has an abundance of structural funct ions in cells. A lot of new classes of non-coding RNAs have been discovered during last decades. For example, the microRNA regulates gene expressio n through post-transcript ional repressio n . The riboswitches are cis-acting regulatory element s, which act as feedback regulator of metabo lite abundance. Bio logical functions of majorit y funct ional RNAs are crucially depend on a secondary structure, which is the scaffo ld of a tertiary structure. Predict ion of RNA secondary structure can be done by minimum free energy (MFE) approach which based on dynamic programming. Measure of the RNA secondary structure significance partially reflects potential to perform cellular function: there were noticed that non-coding RNAs have less free energy than rando m sequences (Clote et al, 2005). Formally, structural potential of a sequence s determines as (Washiet l et al, 2005) , where E(s) is the minimum free energy, and are the mean and the

standard deviat ion of MFE of the set of shuffled sequences with preserved average dinucleotide content. Maintenance of dinucleotide content is important due to stacking interact ions of base pairs. For the given geno me S of the size N, the surface of structural potential determines as matrix o f Z-scores: }, where Further, motif is locally-optimal if it is a local negative peak of the surface.


Here we present program RNASurface that allows fast reconstruction of the surface of structural potential wit h simultaneous ident ificat ion of locally-optimal motifs. An applicat ion to Bacillus subt ilis demo nstrates that this approach better demarcates non-coding RNA from rando m decoy than RNALfold and other window -based approaches, while preserving time and space consumpt ion. Non-coding RNAs have much more subt le signal than protein-coding genes, thus energy-based approaches are inappropriate as ncRNA-finder tools. However, highly accurate determinat ion of structured intervals is useful for several purposes: de novo regulatory and non-coding RNA search preprocessing accurate ncRNA bound definit io n correlat ion of geno me-wide dataset of structural potential with another geno mic features (such as gene bounds, ribosome profiling, transcriptome data)

1. Clote P, Ferre' F, Kranakis E, Krizanc D (2005) Structural RNA has lower free energy than rando m RNA of the same dinucleotide frequency, RNA, 11:578-591. 2. Washiet l S, Hofacker I, Stadler P (2005) Fast and reliable predict ion of noncoding RNAs, Proc Natl Acad Sci USA, 102(7):2454-2459.