Sombrero: Integrating self-organizing neural networks in the search for dna binding motifs

Identification of the short DNA sequence motifs that serve as binding domains for transcription factors continues to be a challenging problem in computational biology. Currently popular methods of motif discovery are based on unsupervised techniques from the statistical learning theory literature. We present here a working prototype of a neural networks based system that aims to tackle the DNA regulatory motif identification problem. The system consists of three modules, the core module being a SOM-based motif-finder named SOMBRERO. The motif-finder is integrated in the prototype with a SOM-based pre-processing method that initialises SOMBRERO with relevant biological knowledge, as well as a self-organizing tree method that helps the user to interpret SOMBRERO's results. The system is demonstrated here using various datasets.