Machine Learning-Enhanced Coarse-Grained Modelling of Telomeric G-Quadruplex Multimers: A Multiscale Study

forthcoming

MCG-QUAD is an integrative study of G-quadruplex (G4) multimers—noncanonical DNA structures—providing a framework to interpret experimental data and connect a microscopic view to macroscopic observables through a novel multiscale in silico approach. G4s are ubiquitous in the genomes of higher eukaryotes and are believed to play key roles in various biological processes. The presence of G4s in the telomeric region has been shown to inhibit telomerase, opening the possibility for G4-stabilizing compounds to be used as anticancer medications. Sequences that form G4s exhibit long folding timescales. G4s are highly polymorphic structures with long-living quasi-stable topologies, a complexity further compounded in multimers. Structural information on G4 multimers is limited, and existing data often lack physiological context. A novel computational approach is necessary to address open questions about the folding/unfolding dynamics of telomeric sequences into G4 multimers, their structure, topological phase space, the role of crowders, and the effects of ligands on G4 formation, stabilization, and properties. Through a combination of classical bottom-up coarse-graining and recent developments in machine learning, MCG-QUAD provides scalable and accurate models that enable bulk, long-timescale simulations. These simulations will be combined with in vitro observables to elucidate previously inaccessible phenomenology, forming an integrative study of G4 multimers in crowded environments. Furthermore, MCG-QUAD explores the interaction of G4 multimers with ligands to inform the design of new anticancer drugs, pushing beyond the limitations of existing in vitro and in silico studies. In collaboration with leading experts and institutions, MCG-QUAD will generate knowledge for the development of effective G4-targeting therapies and a broader understanding of biomolecular behaviour in complex environments, extending beyond cancer research.