MEMOFinder: combining de novo motif prediction methods with a database of known motifs

Abstract

*Background:*

Methods for finding overrepresented sequence motifs are useful in several key areas of computational biology. They aim at detecting very weak signals responsible for biological processes requiring robust sequence identification like transcription-factor binding to DNA or docking sites in proteins. Currently, general performance of the model-based motif-finding methods is unsatisfactory; however, different methods are successful in different cases. This leads to the practical problem of combining results of different motif-finding tools, taking into account current knowledge collected in motif databases.

*Results:*

We propose a new complete service allowing researchers to submit their sequences for analysis by four different motif-finding methods for clustering and comparison with a reference motif database. It is tailored for regulatory motif detection, however it allows for substantial amount of configuration regarding sequence background, motif database and parameters for motif-finding methods.*Availability:* The method is available online as a webserver at: http://bioputer.mimuw.edu.pl/software/mmf/. In addition, the source code is released on a GNU General Public License.