Significance of Knotted Structures for Function of Proteins and Nucleic Acids
Poster Session I
9 – POS
Board 9
Sequence-dependent Potentials of Mean Force and Fragmentary Experimental Data in
Modeling the Spatial Structure of Proteins
Aleksandra Dawid
, Dominik Gront and Andrzej Kolinski
Faculty of Chemistry, University of Warsaw, Poland
Since the first complete crystallization of the protein structure have passed more than half a
century. During this time, both experimentalists and theorists, were focused on discovering the
secrets of evolution. Their work has provided valuable knowledge, but the heart of the matter
remains unexplained. This is a serious problem whose solution would be to drive the
development of many areas of life. The process of proteins folding is highly complex and
depends on a number of parameters. No universal factor initiating the process and controlling its
proper course. A huge number of physico- chemical interactions of short and long-range
determines the precise position of atoms in space. Taking into account the numerous technical
difficulties, cost and effort associated with the experimental assignment of the structure of
proteins and the large scale of the problem, it is necessary to improve automated methods of
theoretical modeling.
The growing content of the PDB database encouraged us to once again analyze these spatial
distributions. Our analysis, conducted for planar and torsional angles as well as for local
distances between the residues may be valuable for deriving the potentials of mean force, which
are useful in the prediction of secondary structure and protein folding simulations. Each of them
is a potential of mean force (PMF) based on knowledge, because it was obtained as a result of
the statistical analysis of the local geometry of the main chain of proteins of known structure.
PMF is based on an approximated probability distribution function along a coordinate, which is
derived from a Boltzmann weighted average. It is useful to know how the free energy changes as
a function of reaction coordinates, such as the distance between two atoms or the torsion angle of
a bond in a molecule. There are two groups of new potentials in the form of the kernel density
estimation (KDE), intended for simulating models with varying degrees of complexity. For the
coarse-grained representation of CABS model there are three potentials: the distance - R15,
plane angle - A13, torsion angle - T14. For the coarse-grained model with full backbone there is
a potential of torsion angle Phi-Psi.
The specificity of the obtained potentials was assessed as results of isothermal simulation using
the new force field and the dynamics of the Monte Carlo. Also, the folding process of proteins,
which come from benchmark, was simulated with a de novo method for Replica Exchange
Monte Carlo. Our ultimate goal is to create a multiscale protocol to obtain the spatial structure of
proteins, so we have both PMFs for full atom and coarse grained.
- 58 -
