Abstract:

The extract of meaningful and accurate information from the wide data sets is an important element in bioinformatics studies. One of the most important challenges faced is the detection of genomic indicators associated with cancer. The numerization of the genome ranges used to solve this problem and the extract of the properties from the ranges are very effective in solving the problem. There are several methods in literature that are used to numerize DNA ranges. In particular, in previous studies, certain statistical parameters are calculated and a distinction is made through these parameters. Also, the calculated parameters are selected based on the experience of the expert. In the approach proposed in this study, entropic-based numeric maping, a new method of maping, transformed DNA sequencing into numeric signals and subsequently digitalized DNA sequencing with the use of Evolutionary Neural Networks (ESA). In the process of identification using the ESA models, the system extracts identification from the data itself. Subsequently, the properties obtained from ESA models were classified by the Support Vector Machine (DVM) and the k-Last Neighbor's Algorithm (k-NN). In this study, a new method is proposed for the classification of breast cancer and healthy genetic sorting groups from DNA series using both the above approaches. The classification accuracy achieved by the recommended method is 85.97%. The results obtained show that deep learning can be an effective method in applications such as the classification of genes in the genom analysis, the discovery of new genes.

Keywords: