Proteome Informatics Group: Research activities
The Proteome Informatics group is part of the Swiss Institute of Bioinformatics (SIB). The group focuses its activities on the development of software tools and databases for proteomics. Besides teaching activities of the PIG members, in particular within the Master's Degree in Proteo mics and Bioinformatics, three axes of research and services have been pursued since the creation of the SIB. Som e of these activities have been ongoing since 1984.
Inference of regular motifs in biological sequences based on multiple space evolutionary exploration
During evolution, some motifs on biological sequences are conserved because they have a better survival than others that are less useful. Detecting such motifs in silico is relevant because they are likely to be involved in biological processes. A common approach is to use multiple alignment or motif discovering methods. However, due to the high combinatory, classical existing methods fail on difficult cases. Our research focuses on optimisation algorithms, dedicated to perform in depth explorations on motif discovering challenges. The problem can be described as follows: given a set of sequences known to be related, extract one word per sequence, in a manner that all extracted words are as similar as possible. As a similarity measure, we use the Kullback-Leibler distance, also known as the relative entropy. We approach the problem from the point of view of the sampling of a search space, by using several neighbourhood-based metaheuristics such as evolutionary algorithms, simulated annealing or hill climbing. However, these methods are not sufficient by themselves. Our key ideas are: 1) using an efficient representation of the search space, 2) combining and adapting these optimisation methods in a hybrid algorithm in order to fit the specific problem, 3) developing new approaches on metaheuristics.