Development of new approaches based on signal processing for disease diagnosis from DNA sequences

Recently, studies with DNA in bioinformatics fields has been quite important in terms of understanding of DNA structure and determining of genes' effects in living organisms. Especially, several studies are carried out related to determining exon regions in DNA sequences, detection of genetic diseases, and comparison of genomic features. In this thesis study, new approaches have been developed that will able to offer optimum solution suggestions intended for diagnosis of genetic diseases from DNA sequences. For this purpose, a comprehensive 3 stage study was completed. a) In the first stage, for the digitization of DNA sequences the performance of the existing mapping techniques was examined comparatively, the highest performing mapping techniques were identified in the result of applications intended for determining exon regions. b) In the second stage, a new numerical mapping technique was developed to digitize DNA sequences, and the performance of this technique was compared with other existing numerical mapping techniques in literature in the different applications. c) In the third stage, DNA sequences that belong to Beta talasemi patients and healthy individuals were digitized by the proposed numerical mapping technique. Feature selection techniques based on statistical metric were used to obtain the feature exaction set from this data. The feature vectors that generated from this extracted information were classified by YSA, and the diagnosis accuracy with 96% was obtained. In the thesis study, the novel entropy-based mapping technique which has been put forward, and the performance in the implemented applications of the disease diagnosis method from DNA sequences which has been improved based on signal processing and artificial intelligence will guide to new studies to do in this field.