Regensburg 2016 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

BP: Fachverband Biologische Physik

BP 16: Posters - Computational Biophysics

BP 16.1: Poster

Montag, 7. März 2016, 17:30–19:30, Poster C

Contact- and distance-based principal component analysis of protein dynamics — •Matthias Ernst and Gerhard Stock — University of Freiburg, 79104 Freiburg, Germany

To describe and understand protein dynamics, systematic dimensionality reduction is crucial. This can be accomplished by principal component analysis (PCA), a linear transformation which removes linear correlations of the coordinates by diagonalizing their covariance matrix. Different types of input coordinates can be used, like dihedral angles (dPCA[1]) or various kinds of distances (e.g. conPCA[2]) or cartesian atomic coordinates. Internal coordinates often provides higher resolution, especially for large-amplitude motion as found in folding systems[3]. In contrast to dihedral angles which mainly reflect the behaviour of neighbouring residues in a protein, distances between pairs of atoms also incorporate information about residues further apart in the primary sequence.

We employ PCA and classify the results based on distances between Ca atoms as well as distances between different residues (including side chains) for various types of well-known model problems, like folding of villin headpiece or functional dynamics of BPTI or lysozyme. We show that it can be advantageous to include only a selected set of coordinates for a PCA because the selection of input variables strongly influences the results of a PCA.

[1] Y. Mu, P. H. Nguyen, and G. Stock, Proteins 2005, 58, 45.

[2] M. Ernst, F. Sittel and G. Stock, submitted.

[3] F. Sittel, A. Jain and G. Stock, J. Chem. Phys. 2014, 141, 014111.