Building innovative representations of DNA sequence facilitate gene finding

Indices that can discriminate DNA sequences´ coding and noncoding regions are crucial elements of a successful gene identification algorithm. Multiscale analysis of various species´ genome sequences facilitates construction of novel codon indices. We´ve developed two novel DNA codon indices, one based on recurrence time and one based on fractal deviation. Because both work well on short DNA sequences, they both hold the promise of being integrated into and thus improving existing gene identification algorithms.