ELL - Energy Landscape Library

ELL - 3.0.0

This document provides reference information about the Energy Landscape Library.

The goal

This C++ programming library provides a platform for generic algorithms to study kinetics and structure of energy landscapes with descrete states. This algorithms need an abstract representation of this states to be applied to a multitude of models and the corresponding energy or fitness landscapes.

Features

Generic algorithms

A set of walks to traverse energy landscapes:

random walks
adaptive walks
gradient (steepest descent) walks
Monte Carlo walks with Metropolis criterion
low energy path calculation for RNA landscapes using the heuristic by Morgan and Higgs

A walk based barrier estimation in the landscape

a direct walk based estimation using pairs of minima (e.g. Morgan-Higgs heuristic for RNA landscapes)
an up- and down-walk based estimation starting in a known minimum of landscape and ending in another one. If this is unknown it can be added.

Exhaustive landscape flooding:

flooding of single basins (standard via hash and a more memory efficient version)
flooding of multiple basins in parallel (in preparation)

All algorithms are fully parameterisable and very flexible concerning output and results. So they can be set up to:

write all visited states to stream
collect them in listings (also compressed possible)
calculate the canonical partition function of the visited states (e.g. of a gradient basin of a minimum)
perform any wanted operation by overwriting the corresponding reporter classes

States = Landscape Models

Lattice protein representations with Pull-Move and Pivot-Move neighborhood for different lattice models with arbitrary distance based energy functions (based on the BIU library)
an RNA representation interfacing the Vienna RNA package for energy calculation
State for Ising spin glass models (ring of spins)
State for the Number Partitioning Problem (ring of spins)
A generic State implementation based on a node-labeled graph. Using this class any arbitrary energy landscape can be encoded and directly used within the ELL. Each node represents a State and is labeled with its energy while the edges encode the neighborhood relation of the energy landscape.

Landscape Topology Representations

To allow for a coarse grained landscape topology representation of minima and additional information as barriers, saddle points, etc. we provide a set of representations via subclassing of LandscapeTopology:

set of minima
barrier tree
saddle network
funnel partitioning (in preparation)

General features

compressed State representations to minimize the memory usage for storing huge numbers of states (e.g. for energy landscape flooding)
generic neighborhood generators
the random number generation can be globally set or one of the already available generators of the BIU library can be used
tests for most of the given functionality to enable a high standard of the source code. Furthermore, these tests can serve as an example for the own first steps using the ELL.

Dependencies

The ELL depends on the following libraries:

the Bioinformatic Utility library (BIU)
the Vienna RNA package

Appendix

Additionally, the documentation also features the following parts:

The following lists and indices are available