Invited Talk (Chemistry)

New trends in simulation of biomacromolecules and nanostructures.

RNDr. Tibor Kožár, CSc. Department of Biophysics, Institute of Experimental Physics,
Slovak Academy of Sciences, Košice, Slovakia


Computer simulations of large biomacromolecular systems and nanoparticles on atomistic level are computationally demanding tasks considering the number of atoms involved in such systems. Proteins, for example, are built from 20 different amino acids(about 9 to 30 atoms each), thus composed from few hundred building blocks. DNA is built only from 4 bases but the chromosome can contain as much as 109 nucleotides. The estimated computer time required for energy calculation during a molecular dynamics simulation increases by the third power of number of atoms under study. If discrete solvation models are considered, the number of atoms in simulation increases even more. For realistic molecular modeling, the length of simulation should allow to monitor conformational transitions present under real experimental conditions. Thus, as the time scale of simulation (in nanoseconds) is an additional factor, powerful computational resources are necessary. As the number of atoms that can easily extend one million (e.g. macromolecules of biological importance such as viruses, surrounded with thousands of water molecules in the simulated solvent box) in addition to the ns length of the simulation, it is clear that powerful technologies are required in order to complete such calculations. There are two options to speed up the simulations: development and use of new methodologies (e.g. coarse-grained or mesoscopic simulations) or application of new hardware and software technologies such as GPU (Graphics Processing Unit) CUDA (Compute Unified Device Architecture) computing. Several programs for molecular modeling were already adapted for GPU computing, e.g. NAMD/VMD, GROMACS, HEX, etc. and some of these high-performance modeling tools will be discussed in our presentation. In addition, insight will be also given into the utilization and performance of methods designated for high-throughput (e.g. molecular docking) biomolecular calculations.