The analysis of k-tuple distribution in the eukaryotic promoter database using the winner-take-all system

One of possible way to identify a putative functional motif in genomic DNA is based on the identification of statistically relevant features on primary sequence. This paper presents a new statistical approach that involve an Unsupervised Neural Classifier. This methodology appears capable of extracting statistically relevant information from the sequence. In this paper we test this methodology on a set taken from the Eukaryotic Promoter Database