Fuzzy attributes of a DNA complex: development of a fuzzy inference engine for codon-"junk" codon delineation

The present study is concerned with the need that exists in bioinformatics to identify and delineate overlapping codon and noncodon structures in a deoxyribonucleic acid (DNA) complex so as to ascertain the boundary of separation between them. Codons refer to those parts in a DNA complex encoded towards forming a desired set of proteins. Also coexist in the DNA structure noncodons (or "junk" codons), whose functions are not so well defined. Such codon and noncodon parts (at least over some sections of a DNA chain) may conform to diffused (overlapping) states exhibiting sharpless boundaries with indistinctive statistics of occurrence of their constituents. Such overlapping mix of codon and noncodon entities constitutes a (fuzzy) universe with information constituent having a fuzzy structure, which can only be identified in descriptive norms with characteristic membership of belonging to certain attributes. Hence, this work is directed to develop a fuzzy inference engine (FIE), which delineates the fuzzy codon-noncodon parts.

Relevant algorithms developed for the fuzzy inference in question are based on information-theoretic (IT) considerations applied to symbolic as well as binary sequence data representing the DNA. Pseudocodes, as needed are furnished.

Simulated studies using human and other bacterial codon statistics are presented to illustrate the efficacy of the approach pursued. The outcome of the study is illustrated via tabulated results and graphs depicting the delineation sought.

The results signify the success of IT-approach pursued in delineating imprecise codon/noncodon boundaries. The FIE applies both for human and bacterial codon statistics.