Various Defuzzification Methods on DNA Similarity Matching Using Fuzzy Inference System

Abstract

A DNA similarity matching using fuzzy inference system is proposed to measure a similarity between human STR (Short Tandem Repeat) based DNA (Deoxyribonucleic Acid) profiles. Moreover, various defuzzification methods are also tested to observe their behavior on different DNA data characteristics. Experiment on real human STR based DNA profile data shows that the proposed DNA similarity matching produces more realistic similarity values compared to those of the conventional one. Experiment on various defuzzification methods on DNA similarity matching shows that Sugeno defuzzification method is more suitable than those of other defuzzification methods.

Family relation and STR-DNA matching using fuzzy inference

“…Fuzzy inference is a process of obtaining new knowledge through existing knowledge using fuzzy logic [18]- [20]. The fuzzy rules are applied are as follows [21] shown in Figure 4. The weighted average method chosen for the calculations is quite easy and the number of degree of similarity only slightly.…”

A Novel Human STR Similarity Method using Cascade Statistical Fuzzy Rules with Tribal Information Inference

“…Previous research has been undertaken to harness the flexibility of Fuzzy Set to the DNA (Deoxyribonucleic Acid) Profile matching's problem. M. R. Widyanto et al [4] proposed a method to assign fuzzy similarity measure [10] between two allele markers, in contrast to crisp 0 or 1 similarity measure used in conventional STR based DNA profile matching algorithm [11]. The need to have a fuzzy similarity measure is triggered by the fact that STR profiles often showed real-valued numbers as an allele marker value instead of natural numbers.…”

“…Value of every DNA loci is represented by fuzzy number (as shown by Figure 2) where the fuzziness value is set to be 0.4 through experiments [4] and the center of the fuzziness is the value of the corresponding loci. The similarity value [10] between an allele of DNA profile evidence and DNA profile reference is given by where is the value of the j-th loci of the DNA profile evidence and is the value of the j-th loci of the DNA profile reference.…”

“…The hypothesis in embedding the tribal information inference in the proposed FIS is to somehow incorporate the statistical property of the tribal characteristic so that the tribal information is not lost in the STR similarity calculation. It has been noted that although M. R. Widyanto et al [4] were adequate in assigning a fuzzy similarity measure between two alleles, the statistical information regarding the two profile's tribal information is lost. This lost information is what the system is trying to regain.…”

“…This paper proposes a novel method to infer the similarity of tribal information using statistical fuzzy rules [7] to deal with uncertainty and impreciseness of STR. This paper is improvement of [4] to include tribal information inference on STR-Based DNA similarity matching using statistical fuzzy rules where the allele marker's statistical distribution probability density function is used as the membership function in the fuzzy rules. The proposed method consists of two cascade Fuzzy Inference Systems (FIS).…”