CLASSIFICATION OF SODAR DATA BY DNA COMPUTING

Abstract

In this paper, we propose a wet lab algorithm for classification of SODAR data by DNA computing. The concept of DNA computing is essentially exploited to generate the classifier algorithm in the wet lab. The classifier is based on a new concept of similarity-based fuzzy reasoning suitable for wet lab implementation. This new concept of similarity-based fuzzy reasoning is different from conventional approach to fuzzy reasoning based on similarity measure and also replaces the logical aspect of classical fuzzy reasoning by DNA chemistry. Thus, we add a new dimension to the existing forms of fuzzy reasoning by bringing it down to nanoscale. We exploit the concept of massive parallelism of DNA computing by designing this new classifier in the wet lab. This newly designed classifier is very much generalized in nature and apart from SODAR data, this methodology can be applied to other types of data also. To achieve our goal we first fuzzify the given SODAR data in a form of synthetic DNA sequence which is called fuzzy DNA and which handles the vague concept of human reasoning. In the present approach, we can avoid the tedious choice of a suitable implication operator (for a particular operation) necessary for the classical approach to fuzzy reasoning based on fuzzy logic. We adopt the basic notion of DNA computing based on standard DNA operations. We consider double stranded DNA sequences, whereas, most of the existing models of DNA computation are based on single stranded DNA sequences. In the present model, we consider double stranded DNA sequences with a specific aim of measuring similarity between two DNA sequences. Such similarity measure is essential for designing the classifier in the wet lab. Note that, we have developed a completely new measure of similarity based on base pair difference which is absolutely different from the existing measure of similarity and which is very much suitable for expert system approach to classifier design, using DNA computing. In the present model of DNA computing, the end result of the wet lab algorithm produces multi valued status which can be linguistically interpreted to match the perception of an expert.