Parts | Days | Selection | Search | Updates | Downloads | Help

BP: Fachverband Biologische Physik

BP 3: Protein Structure and Dynamics

BP 3.2: Talk

Monday, March 7, 2016, 10:00–10:15, H44

Robust Density-Based Clustering to Identify Metastable Conformational States of Proteins — •Florian Sittel and Gerhard Stock — Biomolekulare Dynamik, Physik, Uni Freiburg

Molecular dynamics (MD) simulations of proteins nowadays deliver an extensive amount of data describing dynamics up to millisecond timescales. Recently, Markov state models (MSM) have been recognized as a concise yet valid manner, to describe the relevant dynamics of proteins.

Here, we provide a novel, self-consistent and robust workflow to construct MSMs from the raw data of MD trajectories. This workflow involves (I) the reduction of dimensionality with suitable methods like Dihedral angle Principal Component Analysis, (II) the construction of geometrically defined microstates by a newly developed density-based clustering algorithm, (III) and the construction of the final MSM employing a dynamic clustering algorithm.

Additionally, we introduce a novel type of diagram to easily compare metastable protein states by their respective dihedral angle content.