Geometric approach to string analysis for biosequence classification

Abstract

SummaryTools that effectively analyze and compare sequences are of great importance in various areas of applied computational research, especially in the framework of molecular biology. In the present paper, we introduce simple geometric criteria based on the notion of string linearity and use them to compare DNA sequences of various organisms, as well as to distinguish them from random sequences. Several other theoretical and statistical results are outlined as well. Our experiments reveal a substantial difference between biosequences and random sequences -the former having much higher deviation from linearity than the latter -as well as a general trend of increasing deviation from linearity between primitive and biologically complex organisms.

Geometric Approach to Biosequence Analysis

“…Due to the motivation given earlier that fluctuations in the linearity measures first disappear in samples of length 50,000, we randomly selected substrings of length 50,000 from the larger excerpts of genomes. Random sequences have been generated with independent symbols with uniform distribution (25% each) (see [3] for more details). Given such a string, using elementary techniques we compute and store an array of distances from points of a monotone path representing a biosequence to the corresponding straight line.…”

“…In the framework of this project we have obtained several other related theoretical and experimental results, which are not presented here because of the page limit. Some of these, together with a more detailed description of the experiments outlined in this article are available in a technical report [3].…”