DescriptionThe genome of many viruses is represented by a long single-stranded ribonucleic acid (RNA) molecule that appears to fold into a highly compact organized structure inside the viral shell. Such structure contains a variety of topological motifs, such as hairpins, bulges, multi-loops, and notably RNA pseudoknots. RNA pseudoknots play an important role also in natural RNAs for structural, regulatory and catalytic functions in various biological processes. In particular, it has been recently recognized an interesting interplay between the shape, structure and assembly of icosahedral viral capsids, and the compact RNA packaging topology. The topology of RNA pseudoknots can be effectively studied by using Random Matrix Theory (RMT), by exploiting a correspondence between a graphical representation of RNA structures with pseudoknots and Feynman diagrams of a particular field theory of large random matrices. The theoretical framework of RMT provides a natural analytic tool for the prediction and classification of pseudoknots, since all Feynman diagrams can be organized in a mathematical series, called topological expansion. The PI is interested in studying numerically some recent matrix models based on RMT to describe the structure of viral RNA encapsidated in a viral icosahedral shell. The PI has long experience in the application of RMT to the study of RNA pseudoknots with RMT, as well as on the simulation of the geometry and shapes of icosahedral shells. The simulations the PI intends to perform on XSEDE are Monte Carlo runs of large stochastic matrices, since the matrix model is naturally formulated as zero-dimensional SU(N) field theory of Hermitian matrices. The number of matrices L is equal to number of nucleotides of the RNA molecules, which in viral RNAs can be of the order of L~10^3. Past preliminary studies showed that the size N of the matrices should be sufficiently large to appreciate topological corrections of the order 1/N^2 and 1/N^4 (at least), which implies the simulation of Hermitian random matrices of order N~24 or N~32. Since the number of degrees of freedom for each matrix is N^2~1024, the configuration space has L*N^2~ 10^6 degrees of freedom. While matrix multiplication can benefit of parallel computing capabilities, the need of performing Monte Carlo simulations orients the PI to request High Throughput Computing resources for this initial XSEDE Startup application. Such initial experience will provide the PI a baseline to evaluate the possibility to steer future versions of the code towards HPC capabilities, including GPU or CPU-GPU clusters. Current local computational capabilities are sufficient for developing the codes and running toy-model simulations (N~4), but do not satisfy the PI’s needs for research purposes of large realistic systems. Therefore, XSEDE startup resources are requested to test larger systems, optimize the code and explore code’s scalability, as well as familiarize with the XSEDE platform. (1 row)
OrganizationSiena College
DepartmentPhysics and Astronomy
Sponsor Campus GridOSG-XSEDE
Principal Investigator
Graziano Vernizzi
Field Of ScienceMolecular and Structural Biosciences