Structure Prediction: Homology Modeling

When a sequence of unknown structure has a clear homologous of known structure, it can be modeled based on this structure. If the homology is high and the alignment is good, the core regions of the protein are ussually well modeled. The loop regions and lateral chains of the aminoacids are usually much more difficult to model.

• SWISS-MODEL. [Access]

There is a lot of homology modeling programs. Unfortunatelly, they use to be difficult to use and require previous expert knowledge. In contrast, SWISS-MODEL has a easy-to-use WWW interface, is fully automatic and its acuracy is comparable to the most complex programs.

• Main problems: Rotamers and loop modeling.

The most important limitatios to the homology modeling are, on the one hand, to predict the different spatial disposition of the lateral chains of the peptide residues; and on the other hand, the spatial prediction of the loops. This last problem is very difficult to resolve.
As a example, here are shown:
• 1.- A figure with the spatial disposition of the lateral chains of ten aminoacids shared by different structures. [HERE]
• 2.- A figure showing the superposition of three loop structures: the template, the model and the actual structure. [HERE]

Model evaluation.

• Biotech  Validation Suite for Protein Structures. [Information]

Server that evaluates the quality odf the structure using 3 different programs and combining the results.
•  Checks performed by PROCHECK V3.5:

     Covalent geometry
     Planarity
     Dihedral angles
     Chirality
     Non-bonded interactions
     Main-chain hydrogen bonds
     Disulphide bonds
     Stereochemical parameters
     Parameter comparisons
     Residue-by-residue analysis
• Checks performed by PROVE V2.3:

     The Atomic Volume Check
• Checks performed by WHAT IF V4.99:

     Verification of bond angles
     Verification of bond lengths
     Buried Hydrogen Bond Donor Check
     Bump Check
     Verification of chain names
     Peptide Plane Flip Check
     Chiral Handedness Check
     HIS GLN ASN side chain conformation Check
     Nomenclature Check
     Check of side chain planarity
     Check of proline puckering
     Quality Check
     Side-chain Rotamer Check
     Symmetry Check
     Torsion angle Check
     Check of water clusters
     Check of atomic occupancy
 
• Solv_Pref.

Program by L. Holm y C. Sander that calculates the solvent exposition of the model residues and compare them with residue solvatation preferences from databases.
 
 
• ProSA (Protein Structure Energetic Analysis). [Information](M. Sippl)

Program based on energy potentials extracted from databases. It evalues the model energy based on this potentials. Is believed that a bad model gives a higher energy value that a good one.
•     Energy analysis of oligomers and multi-chain complexes.
•     Frameshift error detection in electron density interpretation.
•     Sequence structure alignment and fold recognition techniques.
•     Access to Prosa II's energy functions.

PRACTICALS: