Description of LINUS

Introduction

LINUS (Local Independently Nucleated Units of Structure) is a program to simulate protein folding. At its core LINUS is a Metropolis Monte Carlo procedure. As with any molecular simulation program, at least 3 decisions have to be made.

Representation of the Protein Structure
The procedure for exploring conformation space
The energy function to use.

Representation of the Protein

LINUS represents a protein by including all it's heavy atoms (i.e., non-hydrogen atoms). No hydrogens of any kind, polar or non-polar, are included. A structure suitable for simulation with LINUS can be generated from a sequence file or a Brookhaven PDB format file using programs included in the LINUS distribution.

Exploring Conformation Space

To explore conformational space, LINUS uses a move set. The moves in LINUS operate on contiguous 3 residue segments. Four different types of moves are defined.

α-helix move
β-strand move
β-turn move
coil move

Details of the torsion angles for each of the individual moves are available by following the Move Set link in the side bar.

The choice of a 3 residue segment was arrived at by recognizing that setting 3 consecutive residues to an α-helical conformation guarantees the formation of one helical hydrogen bond - thus effective abolishing the helix initiation factor in traditional helix-coil theories.

Energy Function

The LINUS energy function is more appropriately referred to as a scoring function. Traditionally energy functions fall in one of two classes - molecular mechanics based force fields and knowledge potentials derived from analysis of experimentally known protein structures. The LINUS scoring function is based on generally known characteristics of protein structures. Thus, it is known that in globular proteins:

the backbone peptide groups participate extensively in inter-residue hydrogen bonding
the interior of the protein is enriched in non-polar groups

LINUS makes qualitative use of these observations in its scoring functions. The four components of the scoring function are:

Hard Sphere Repulsion - no two atoms can be at the same place at the same time
A contact energy term that rewards bringing hydrophobic residues into proximity
A hydrogen bond term that rewards the presence of hydrogen bonds between backbone atoms and also between backbone donors and sidechain acceptors
A torsion term. This term rewards glycine and asparagine if the φ value is positive and penalizes other residues if they adopt a positive φ value.

In addition to the generic hydrogen bond term it is possible to check for the absence of hydrogen bond satisfaction. If a backbone N or O is not hydrogen bonded intramolecularly or accessible to solvent water, a penalty is assessed.

Finally, this version of LINUS includes a function to estimate the solvation energy of a protein conformation using the Conditional Accessible Surface Area (CHASA) algorithm described by Fleming and Rose (2005).