A Novel Computational Method for Characterizing and Analyzing Genomic Sequences with Applications to Phylogenetic Analysis for SARS-Associated Coronavirus

摘要

As it is known, the SARS virus is mutating very rapidly. This made diagnosis of related diseases and the discovery of vaccines against said virus more difficult. In recent articles in science magazine, the SARS virus was sequenced and analyzed. In these articles, the computational methods used to compare SARS to other coronaviri were based on statistical techniques. In this paper we discuss the limitations of such statistical methods and the need for more accurate and computationally efficient methods to characterize the SARS among other coronaviri. In particular, in statistical methods important information, such as the location of blocks of similarities in a given sequence, is not a factor taken into account in the calculation of global sequence similarity scores. We present here a novel technique to uniquely and compactly represent the SARS and the other coronavirus sequences. In a previous work through a mapping process, we demonstrated the ability to 1) recognizes shapes, and 2) concisely represent shape of large data using a set of coefficients derived in the mapping process. In [13-17] we have demonstrated how this method was applied to fingerprint, object representation and recognition, facial and large data representation and recognition. In this work we illustrate how these previous results can be applied to DNA sequence data Phylogenetic and similarity analysis. In the approach outlined herein, a syntactic representation of DNA sequences is formed as polygonal shapes from which we extract a compact set of coefficients uniquely representing the DNA sequence. Additionally, we present and show how the size of the sequence data does not affect the computational performance of the proposed technique. We also show how the proposed compact sequence representation is highly sensitive to minor changes such as a single mutation or SNP change, which makes the proposed technique very accurate in comparison to current statistically, based methods. Additionally, we show how the proposed technique allows the effect of "gaps" between the aligned matching blocks to be taken into account. More importantly, we show how adding the information, such as the nucleotide positions and the spacing between these blocks affect the similarity score. This information is very interesting for the phylogenic studies of genes, and organisms. To the best of our knowledge there has been no work in the literature where the information related to the "gaps" and their positions between matching blocks in sequences has been involved in the similarity scores. Finally, experimental results using several coronaviri are presented to show the potential value and power of the proposed technique in placing the SARS virus among the coronaviri and precisely tracking its rapid mutations.