Year of Graduation
Making Sense of Genomic Data with Machine Learning Methods
Applied Mathematics and Information Science
After the discovery by J. Watson and F. Crick of the canonical right-wound form of the double helix DNA, called the B-form, it was proved that there are other DNA conformations and secondary structures such as A-form and Z-form, quadruplexes, and cross-shaped structures. It is known that the secondary structures of DNA (RNA) play an important role in various processes of genome functioning, such as chromatin organization, transcription, translation, and other. While there are many available experimental data on which the genome annotation is based on various genomic elements, high-flux experimental devices have been invented so far, allowing the detection of secondary DNA structures (RNAs) and determining the significance of their role in the genome. This paper consists in the construction of algorithms and methods that allow annotating the secondary structures of DNA (RNA) with known genomic and epigenetic elements. Machine learning methods based on unsupervised learning algorithms, such as clustering and dimensional reduction methods, are used for computer processing and searching for dependencies in the organization of secondary structures in the genome. The final result of the work was the creation of a universal structure for storage and processing of genomic data represented in the form of interval sequences, as well as clustering, processed by such a structure, data.