KOALAB: A new method for regulatory motif search. Illustration on alternative splicing regulation in HIV-1
Abstract
Discovering heterogeneous regulatory motifs remains a difficult problem in biological sequence analysis. In this context, statistical learning or pattern search techniques on their own have shown some limitations. However, significant benefits can be taken from their complementarity. We selected two state-of-the-art methods: a multi-class support vector machine (M-SVM) from the statistical learning domain associated with a performant discrete pattern matching algorithm grappe, and in- tegrated them into a web technology based graphical software: KOALAB (KOupled Algorithmic and Learning Approach for Biology)1 . We applied our method on motif discovery within nucleic acid sequences using experimental SELEX results as training database for the M-SVM. An application dealing with the search for splicing regulatory protein binding sites in HIV-1 genome shows the potential of such an approach.
Loading...