Parts | Days | Selection | Search | Updates | Downloads | Help
BP: Fachverband Biologische Physik
BP 12: Poster II
BP 12.61: Poster
Tuesday, April 2, 2019, 14:00–16:00, Poster B2
Identification of metastable conformational states of protein dynamics — •Daniel Nagel — Biomolecular Dynamics, Institute of Physics, Albert Ludwigs University, 79104 Freiburg, Germany
Well-defined microstates, describing metastable conformational states, are the key to generate a Markov state model of protein dynamics. Taking a dimensional-reduced trajectory, these can be found by a recently proposed density-based geometrical clustering algorithm by Sittel et al., which is self-consistent in its data-based input parameters and computationally efficient. While for simple geometrical clustering methods such as k-means it was necessary to rely on a large number of microstates to properly discretize barriers and subsequently use dynamic clustering techniques to reduce the large set of microstates to a manageable number of macrostates, density-based clustering by design cuts at the energy barriers producing directly a reasonable number of microstates. Even though the latter algorithm performs much better, projection artifacts in the transition regions artificially shorten the estimated lifetimes. A simple corrective is to use dynamical boundary corrections, namely dynamical coring which requires staying for a minimum time after a state transition in the new state for the transition to be counted. This method increases metastability and Markovianity significantly by identifying misclassified interstate fluctuations as intrastate fluctuations. To illustrate the simplicity and performance of the workflow, two well-established biomolecular systems (alanine dipeptide and villin headpiece) are examined.