INFORMATION-CONTENT IN NUCLEOTIDE-SEQUENCES

We assessed the information content in nucleotide sequences through th e efficiency of complete nucleotide sequence reconstruction from a set of its fragments (frequency-correlation dictionary), using the increa se in the reconstructed sequence entropy for frequency-correlation dic tionaries of q-letter-long words as a measure of efficiency. Human gen es have a maximum increase with q = 5, 6, and 7. We also revealed abno rmally efficient reconstruction using dictionaries with q = 3 and 2, w hich distinguishes the natural nucleotide sequences from random ones.