Sequence-structure-function relationships of nucleic acids
Year: 2025
Type: dissertation
Abstract: Structure has been shown to have a significant influence on the biological function of nucleotide sequences. Data from a published Loop-seq experiment, which is a high-throughput method for measuring the intrinsic flexibility of DNA fragments termed 'cyclizability,' was used to develop a deep learning model for predicting DNA flexibility. The model identified a consistent, dinucleotide-driven flexibility pattern near transcription start sites and highlighted associations between DNA flexibility and transcription factor binding, influenced by the binding motifs flanking sequences. Global differences in cyclizability between genomic and randomized sequences were observed, with eukaryotes, excluding vertebrates, showing a preference for rigid sequences. In Arabidopsis thaliana, rigid regions were associated with higher mutation rates, both de novo and fixed. Sequence-structure relationships were further investigated using generative diffusion models for nucleotide sequence design. Diffusion models were applied to generate RNA and DNA sequences based on structural constraints. With a proven performance in tRNA structure prediction we extended our model to generate sequences that conform to specific secondary structures. Additionally, a further model was created to generate DNA sequences with targeted cyclizability values, which was later enhanced to incorporate partial sequences, allowing flexibility tuning while preserving specific motifs. Finally, the role of DNA methylation in metabolic adaptation to cold stress was examined in Arabidopsis thaliana. Epigenome-wide association studies identified novel potential targets involved in metabolic adaptation to low temperatures. Taken together, this thesis explores the mechanical properties of DNA, generative nucleotide sequence design, and the role of DNA methylation in metabolic adaptation to temperature.
Author: Back, Georg
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W6889561126
Cites:
Cited by:
Related to: