DNA Sequence Complexity Reveals Structure Beyond GC Content in Nucleosome Occupancy

We introduce methods that rapidly evaluate a battery of informationtheoretic and algorithmic complexity measures on DNA sequences in application to potential binding sites for nucleosomes. The first application of this new tool demonstrates structure beyond GC content on DNA sequences in the context of nucleosome binding. We tested the measures on well-studied genomic sequences of size 20K and 100K bps. The measures reveal the known in vivo versus in vitro predictive discrepancies, but they also uncover the internal structure of G and C within the nucleosome length, thus disclosing more than simply GC content when one examines alphabet transformations that separate and scramble the GC content signal and the DNA sequence. Most current prediction methods are based upon training (e.g. k-mer discovery), the one here advanced, however, is a training-free approach to investigating informative measures of DNA information content in connection with structural nucleosomic packing.