Institute of Computer Science
University Freiburg

Energy Landscapes

Energy landscape 3D The concept of energy landscapes has proven to be of fundamental relevance in investigations of complex disordered systems, from simple spin glass models to biopolymer folding. In this picture, energy is viewed as an explicit function E(S) of underlying conformational degrees of freedom S. The topological structure of the conformation space is determined in terms of the elementary moves that underly the dynamical behavior. Examples are single spin flips in spin glasses, the formation or breaking of a base pair in RNA folding models, or rotation around a bond in a protein folding model.

The geometric properties and topological details of the energy landscape, such as number of local optima, the saddle points separating them, as well as the size distributions of the basins of attraction, therefore directly influence the dynamics of the underlying system. A thorough understanding of these aspects of geometricall andscape structure is thus of wide interest.

The Energy Landscape Library - a Platform for Generic Algorithms

ELL-logo The ELL-library provides a platform for generic algorithms to study kinetics and structure of energy landscapes with discrete states. These algorithms need an abstract representation of these states to be applied to a multitude of state instances and their corresponding energy or fitness landscapes.
Click here for further details

Coarse Energy Landscape Representations for RNA Molecules

barrier tree Based on the energy landscape library, a set of tools to investigate the structure and topology of RNA energy landscapes has been implemented. These programs allow for:
  • Minima sampling
  • Exact calculation of barrier trees or saddle networks
  • Approximation of barrier trees
  • Barrier estimation using a heuristic by Morgan and Higgs
Click here for further details

Energy Landscapes and Kinetics of Lattice Proteins

barrier tree For this research topic, we analyze the energy landscape of three-dimensional model proteins. We plan to investigate new techniques for the construction of energy landscape representations and the computation of protein kinetics. We are furthermore interested in the analysis of landscapes of even more complex protein models.

Current status

We could show how protein structure prediction helps in the construction of barrier-trees, which represent local minima and energy barriers of the energy landscape. Starting from predicted global optima and structures on the first raised energy levels, we can cover the structure space up to a defined degree.


  • CPSP-tools - Optimal structure prediction etc. in 3D HP models
  • LatPack Global and sequential (vectorial) folding simulation in arbitrary 3D lattice protein models
  • LatFit High resolution fitting of 3D full atom protein data (PDB) onto lattices

Contributing group members

Main Publications