Jump to main content
Jump to site search

Issue 22, 2011
Previous Article Next Article

Which similarity measure is better for analyzing protein structures in a molecular dynamics trajectory?

Author affiliations

Abstract

An important step in the computer simulation of the dynamics of biomolecules is the comparison of structures in a trajectory by exploiting a measure of distance. This allows distinguishing structures which are geometrically similar from those which are different. By analyzing microseconds-long all-atom molecular dynamics simulations of a polypeptide, we find that a distance based on backbone dihedral angles performs very well in distinguishing structures that are kinetically correlated from those that are not, while the widely used Cα root mean square distance performs more poorly. The root mean square difference between contact matrices turns out instead to be the metric providing the highest clustering coefficient, namely, according to this similarity measure, the neighbors of a structure are also, on average, neighbors among themselves. We also propose a combined distance measure which, for the system considered here, performs well both for distinguishing structures which are distant in time and for giving a consistent cluster analysis.

Graphical abstract: Which similarity measure is better for analyzing protein structures in a molecular dynamics trajectory?

Back to tab navigation

Additions and corrections

Publication details

The article was received on 26 Nov 2010, accepted on 10 Mar 2011 and first published on 04 Apr 2011


Article type: Communication
DOI: 10.1039/C0CP02675A
Citation: Phys. Chem. Chem. Phys., 2011,13, 10421-10425
  •   Request permissions

    Which similarity measure is better for analyzing protein structures in a molecular dynamics trajectory?

    P. Cossio, A. Laio and F. Pietrucci, Phys. Chem. Chem. Phys., 2011, 13, 10421
    DOI: 10.1039/C0CP02675A

Search articles by author

Spotlight

Advertisements