Berlin 2018 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
BP: Fachverband Biologische Physik
BP 12: Computational Biophysics I
BP 12.3: Vortrag
Dienstag, 13. März 2018, 10:00–10:15, H 1058
G-PCCA - a generalized Markov state modeling approach for both equilibrium and non-equilibrium systems — •Bernhard Reuter1, Konstantin Fackeldey2,3, Susanna Röblitz2, Marcus Weber2, and Martin E. Garcia1 — 1Theoretical Physics II, University of Kassel, Germany — 2Zuse Institute Berlin (ZIB), Germany — 3Institute of Mathematics, Technical University Berlin, Germany
Markov state models (MSMs) have received an ongoing increase in popularity in recent years as they enable the conflation of data from simulations of different length and the identification and analysis of the relevant metastabilities and kinetics of molecular systems. However, so far methods and tools for building MSMs, like PyEMMA and MSMBuilder, are restricted to equilibrium systems fulfilling the detailed balance condition. This constitutes a severe constraint, since molecular systems out of equilibrium, e.g. disturbed by an external force, have attracted increasing interest. To overcome this limitation we have developed and implemented - in Python and MATLAB - a generalization of the widely used robust Perron cluster cluster analysis (PCCA+) method, termed generalized PCCA (G-PCCA). This method, based on the utilization of Schur vectors instead of eigenvectors, can handle both data from equilibrium and non-equilibrium simulations. G-PCCA is able to identify dominant structures in a more general sense, not limited to the detection of metastable states, unraveling cyclic processes. This is exemplified by application of G-PCCA on non-equilibrium molecular dynamics (NEMD) data of the Amyloid-beta peptide, periodically driven by an oscillating electric field.