Net Nearest Neighbor Analysis (NNNA) summarizes non-compensated dinucleotides within gene sequences

Abstract

Motivation: Net Nearest Neighbor Analysis (NNNA) measures a previously unexamined aspect of dinucleotide frequency - the non-compensated, non-repetitive dinucleotides in a sequence. Non-compensated dinucleotides are those in excess of their corresponding reverse dinucleotides. Results: NNNA regards dinucleotides as vector quantities, making it possible to summarize any sequence as a group of circuits and tags. The results of NNNA are found to be consistent with traditional analytic methods, yet reveal additional characteristics of the sequences. The NNNA circuits and tags uniquely identify each tRNA in Escherichia coli K-12 and certain structural components of each tRNA, extract function-specific characteristics for each of the sequences involved in the formation of insulin from preinsulin, and exhibit species-specific phylogenetic characterization (demonstrated with Monilinia). Availability: Nearest neighbor analysis software has been available for many years and is a component of most gene analysis software packages, including the Staden Package which is available at no charge to academic users (http: //www.mrc-1mb.cam.ac.uk/pubseq/).