We report statistical studies of correlation properties of similar to 7500
gene sequences, covering coding (exon) and non-coding (intron) sequences fo
r DNA and primary amino acid sequences for proteins, across all three domai
ns of life, namely Eukaryotes (cells with nuclei), Prokaryotes (bacteria) a
nd Archaea (archaebacteria). Mutual information function, power spectrum an
d Holder exponent analyses show exons with somewhat greater correlation con
tent than the introns studied. These results are further confirmed with hyp
othesis testing. While similar to 30% of the Eukaryote coding sequences sho
w distinct correlations above noise threshold, this is true for only simila
r to 10% of the Prokaryote and Archaea coding sequences, for protein sequen
ces, we observe correlation lengths similar to that of "random" sequences.
(C) 2000 Elsevier Science B.V. All rights reserved.