Preparing high quality model of a protein structure for
computational chemistry studies starts with addition of hydrogen atoms to PDB coordinates
of heavy atoms implying correct assignment of ionization states of protein residues and
optimization of proton positions, which is a challenging scientific task.
Lead-Finder saves user’s time and reduces expenses by including protein structure preparation
tools that allow the:
- flexible addition of hydrogens to protein
heavy atoms (usually present in PDB files and homology models);
- optimization of polar hydrogen positions with respect to the
ligand, substrate and cofactor present in the structure file;
- assignment of optimal ionization states to protein
residues at a physiologically relevant pH;
- automatic processing of ligands/substrates/cofactors present
in the protein structure file through the addition of all hydrogen atoms
to these molecules;
- optimization of side chain orientations of His, Asn and Gln residues for
which X-ray analysis can return flipped orientations due to apparent symmetry;
- correction of errors and inconsistencies typical of PDB and
other protein structure files;
- automatic building of missing (unresolved) terminal residues and side chains of the protein;
- cleaning structure files of waste molecules (buffer components, water, etc.);
Lead-Finder uses quite sophisticated theoretical approaches to assign optimal ionization states of protein
residues at arbitrary pH, which is based on the recently introduced screened coulomb potential
(SCP) model. SCP theory accounts for the dependence of electrostatic interactions between charged particles
on such physicochemical properties of their microenvironment as hydrophilicity and degree
of solvent exposure. Description of the SCP model can be found in original publications
1,2
, and details of its implementation in Lead-Finder program are given in the Technology section
The quality of Lead-Finder predictions of ionization properties of proteins was validated
by calculating pKa values of 100 residues from 15 proteins, for which robust
experimental data were available. Lead-Finder demonstrated impressive results, yielding RMSD of predicted
vs experimental pKa values of 0.7 for Asp and Glu residues, 0.8 for His residues,
0.85 for Lys residues, 1.25 for Tyr residues. Details of benchmarking experiments can be found
in section Accuracy of pKa predictions.
|