Application of GANs to the Tasks of Recognition of DNA Seocndary Structures

DNA is one of the most important macromolecules in all of biology. It consists of sequences of nucleotides and is responsible for storing and transmitting information as well as for many functions in the cells of organisms. The DNA secondary structure is the three-dimensional form of DNA. It determines the structure and function of the molecule. The most two common structures a A-DNA and B-DNA. There is also Z-DNA, in which the helix twisted to the left, contrast to A and B. The definite laws are currently unknown, and the amount of experimentally obtained data is small. Therefore, the recognition and generation of Z-DNA using neural networks is an important task. This work is devoted to the application of generative adversarial neural networks to the problems of recognizing Z-DNA. The task of recognizing Z-DNA is one of the most important, since its solution makes it possible to determine whether a given nucleotide sequence is a left-twisted DNA helix or not. The task of generation is to create nucleotide sequences in which DNA will have the structure of Z-DNA. Generated sequences can be used in improving the quality of other models, which require a large amount of data. As a result of this work, there were made neural networks, that can recognize Z-DNA with an accuracy of 1.0 and can generate sequences that are recognized by CNN with accuracy of 0.97. In addition, before the recognition and generation of Z-DNA, these neural networks were tested on the promoters of human DNA. This work is one of the first works on the use of neural networks in the generation and recognition of z-DNA