A Convolutional Code-Based Sequence Analysis Model and Its Application

Abstract

A new approach for encoding DNA sequences as input for DNA sequence analysis is proposed using the error correction coding theory of communication engineering. The encoder was designed as a convolutional code model whose generator matrix is designed based on the degeneracy of codons, with a codon treated in the model as an informational unit. The utility of the proposed model was demonstrated through the analysis of twelve prokaryote and nine eukaryote DNA sequences having different GC contents. Distinct differences in code distances were observed near the initiation and termination sites in the open reading frame, which provided a well-regulated characterization of the DNA sequences. Clearly distinguished period-3 features appeared in the coding regions, and the characteristic average code distances of the analyzed sequences were approximately proportional to their GC contents, particularly in the selected prokaryotic organisms, presenting the potential utility as an added taxonomic characteristic for use in studying the relationships of living organisms.

Applications of Convolutional Codes to DNA Codes and Error-Correction

“…The amino acid correspondence table is depicted below. and error-correcting information of adjacent codons [19]. In particular, it has been shown that there is a "strong short-range correlation of adjacent bases" and that adjacent genes are more likely to be co-expressed than non-adjacent genes [20] [11].…”

“…Additionally, the expression patterns of adjacent genes are more correlated than patterns of randomly selected gene pairs [16]. In all, various studies have demonstrated that the nearby informational units have an effect on the current informational unit [19].…”

“…Since initial experiments conducted by May et al in the early 2000s that explored the role of convolutional codes in biocomputation, their relevance has grown as they can take genetic processes into account [23]. While current convolutional code models have taken necessary requirements and characteristics of DNA codes into consideration, various issues still need to be addressed [19]. Understanding how outside influences and biological characteristics can further factor into DNA encoding and its mathematical representation is relevant.…”

Survey on encoding schemes for genomic data representation and feature learning—from signal processing to machine learning

“…For example, in mapping using the binary codes, Hamming distance and Euclidean distance were employed to reflect the differences between transition and transversion in light and dark colors, respectively, as shown in Fig. 3 [2,34,35] .…”

“…A convolutional encoding method was adopted, where the error-correction coding theory was used to encode and analyze DNA sequences for prokaryotic organisms [35] . The convolutional code model was designed based on degeneracy of codons.…”