SEMINAR

Bioinformatics: Shannon’s negentropy perspectives

Perambur S. Neelakanta, Department of Electrical Engineering, College of Engineering and Computer Science, Florida Atlantic University, Boca Raton, Florida 33431

Abstract

The research addresses negentropy aspects of genomic details elucidated in terms of information-theoretic perspectives (due to Claude Shannon).  Information theory is a probability-theoretic approach based on negative-entropy (or non-uncertainty) details buried in a set of entities (or signal elements).The underlying art can be used in rational contexts of perceiving useful or meaningful details from a set of stochastical considerations associated with the subject of interest, such as bioinformatics.

For example, application of information theory (IT) or negentropic concepts to genomic details, concerns with the information stored in nucleotide/DNA sequences and in the translation process leading to molecules of life proteins. The concept of entropy or measure of disorderliness (H), when viewed negatively in  Shannon’s sense leads to the celebrated information metric, I = – H = –k where p_i is the probability associated with the set of stochastical entities involved. A companion formulation depicting the relative entropy (or mutual information) between two sets of probability distributions {p_i} and {q_i} is I_M =

Identified here is a gamut of avenues in bioinformatics where the concept of information theory can be fruitfully applied: Examples presented refer to:(i) The informatic structure of DNA sequence – Redundancy and noisy considerations; (ii) genetic coding: Discrimination/delineation of crisp and/or fuzzy codon-noncodon boundaries; (iii) gene expression – elucidation of specific information such as segregation of promoter section TATA box from putative TATA boxes; (iv) estimating the base compositional symmetry between complementary DNA strands of mitochondrial chromosome; (v) complexity metric versus negentropy – application in bioinformatic sequence discrimination efforts; (vi)  Csiszar’s family of (neg)entropy formulations: Applications to bioinformatics; (vii) tracking HMM in IT-domain: Relevant Viterbi/forward algorithms of dynamic programming in biological pursuits;(viii) phylogenetic predictions – use of various/unusual statistical distance/divergence measures etc.

Contact for more details: rnarayan@fau.edu, URL: http://www.science.fau.edu/fbrcf/