Software: Proteins - Folding

Prediction of protein structure based on sequence information alone is one of the important challenges of contemporary computational biology. The number of protein sequences deposited in genomic databases grows very fast; however the number of known protein structures deposited in the Protein Data Bank (PDB) increases incomparably slower. Since the knowledge of the three-dimensional structure of a protein is often a key to its function and properties, protein structure prediction algorithms and programs are becoming increasingly more important. There are three classes of approach to the structure-prediction problem: 

• Sequence-homology based methods applicable when there are known structures of proteins with high sequence similarity to a protein under study, these methods take advantage of the empirical relationship between sequence and the three-dimensional protein structure.

• Threading-based methods use a database of known three-dimensional protein structures and an empirical scoring function in order to choose which three-dimensional known structure (or structures) is probably closest to the structure of protein under study.

• Ab initio (or de novo) methods use an energy function coupled with a search method to find unknown structure far a given protein.

The software available here covers the last two classes of approaches to the protein structure prediction.

LOOPP (Learning, Observing and Outputting Protein Patterns) is a fold recognition program based on the collection of numerous signals, merging them into a single score, and generating atomic coordinates based on an alignment into a homologue template structure. The signals we are using include straightforward sequence alignment, sequence profile, threading, secondary structure and exposed surface area prediction. This program represents a threading-based approach. The program can be used directly from this web page.

ECEPPAK (Empirical Conformational Energy Program for Peptides Package) contains various modules performing different calculations with protein structure and properties. This package also contains a Monte Carlo Minimization-based search module (Electrostatically-Driven Monte Carlo) for locating low-energy conformations of proteins and polypeptides in all-atom representation with the ECEPP/3 force field.

PROTARCH package uses a united-residue representation of the polypeptide chain and  the Conformational Space Annealing search method to predict a protein structure. The key stage of the method is global optimization of the energy of a polypeptide chain at the simplified united-residue level. PROTARCH also includes computational tools to carry out cluster analysis and construction of all-atom backbone from the united-residue chain using the dipole-path method.