An information theoretic approach for the discovery of irregular and repetitive patterns in genomic data