CeBiTec Colloquium
| date | Monday, July 19th 2010, 17 c.t. | |
| location | G2-104, CeBiTec Building | |
| speaker | Dr. Annelyse ThéveninLaboratoire de Recherche en Informatique, | |
| title | A pseudo-boolean programming approach to compute differences and similarities between two genomes (and to predict RNA structure) | |
| Comparative genomics aims to better understand differences between species. Several methods for genome comparison exist; in this talk, we focus on the computation of three measures of (dis)similarities, namely the number of adjacencies, the number of breakpoints, and the number of common intervals. In presence of duplicated genes or when the order of genes is only partially known, computing these measures is an NP-hard problem. First, we want to compute the number of adjacencies and the number of breakpoints for three models (exemplar, maximum and the intermediate model introduced in this work) between two genomes with duplications. In order to obtain exact results, we use a pseudo-boolean programming approach. After a test on 12 genomes of gamma-proteobacteria, we get enough results to compare different combinations of measure/model. Additionally, we propose and evaluate (thanks to the above-mentioned results) a family of heuristics based on a search of a longest common subsequence, which gave very good results on these data. Second, we set up a pseudo-boolean approach to compute the number of adjacencies and the number of common intervals between two partially ordered genomes. Using nearly 800 simulated genomes, we have studied the influence of parameters associated to partial orders and compared both measures. Finally, we propose a method of prediction of structure of RNA. This preliminary work is based on the comparison of a sequence of RNA with a known structure of a reference RNA. We use a pseudo-boolean programming approach for this prediction. | ||
| host | Prof. Dr. Jens Stoye |