A new approach for mutation analysis using data mining techniques

In this study, a new method is proposed to be used in diagnostic process of genetic disorders to determine the mutations in DNA sequences. The contribution of our method is that it uses chromatograms without applying a base calling method in order to decrease the errors produced during the base calling step. Given reference and unknown chromatograms, our method searches for possible mutations in the unknown chromatogram against the reference one. Our approach first extracts feature vectors of both chromatograms by applying a two dimensional transformation to every data frame sliding through the chromatograms. The feature vectors are then used to obtain similarity matrix proceeded by applying dynamic programming from which differences between them are displayed. Difference plot can be used either for manual screening or automated mutation detection. We test our method on a freely available dataset. The results show that our method can successfully align two chromatograms and highlight the differences caused by mutations.