| Abstract [eng] |
This study on DNA properties of computationally mapped nucleosome positions was done by Indiras Maziukas, Vilnius University, for the systems biology study programme, supervised by doc. dr. Erinija Pranckevičienė. This study aimed to develop a Hidden Markov Model based computational nucleosome mapping tool and integrate it into an existing Galaxy instance, addressing the need for accurate and user-friendly tools in nucleosome position prediction. The HMM-based algorithm was created and trained using known human and mouse nucleosomal sequences and tested on 204 nucleosomal sequences with known nucleosome positions. Custom software was employed for data preparation and analysis, streamlining the workflow with documentation available for its use. The HMM-based tool and additional helper tools were not integrated into the Galaxy Platform, but the platform itself was used. Despite the model's limited accuracy in predicting nucleosome positions when compared to existing literature, valuable insights were gained regarding the challenges and limitations of computational predictions in this field. The study highlights the importance of optimizing parameters and methodologies, expanding the scope of datasets. Overall, this research contributes to the growing body of knowledge in nucleosome position prediction and supports further efforts to understand the complex mechanisms underlying gene regulation and chromatin structure. |
