PDB Preparation for Docking

---

Is the structural model valid?

Before you do any modeling with a protein structure obtained from the PDB it is a good idea to check the validity of the structure. The crucial question is of course whether the model that the crystallographer built is correct. Unfortunately, there is no magic test to answer this question. Instead, here are a number of tools that examine whether the structure is conform with a priori knowledge about protein structures. Typical questions are:

• Does the model fit the electron density?
• Is the model geometry consistent with bond lengths, angles and dihedrals observed in crystal structures of invidual amio acis or small peptides?
• Are there steric clashes in the model?
• Are there uncompensated charged groups or unsatisfied hydrogen bonding functional groups buried in the protein interior?
• Is the crystallographic R-factor sufficiently low given the resolution of the data.
• Does the model contain too many waters?
• Was B-factor refinement warranted?
• Are Asn, Gln and His side chains correctly modeled, or should their planar groups be flipped?
• Should peptide planes be flipped?
• Are waters to be trusted?

Many of these questions can be answered by reading carefully the PDBREPORT on your PDB. Alternatively, you load your PDB into the Biotech Validation Suite for Protein Structures.

How flexible is the protein?

In case you have more than one structure of the same protein it may be possible to gain information on the flexibility of the protein. Typically, you might find evidence of:

• domain motions
• loop movements
• alternative side chain rotamers

The consequences are that it may be necessary to perform docking experiments with more that one protein model. In case of modest changes, such as alternative side chain rotamers, they might be included directly as extra degrees of freedom in the search procedure.

Where are the hydrogens?

In omst cases the resolution of crystal structures is insufficient to localize hydrogen atoms. Hence, their positions have to be dedudced. For many atoms the hybridization and connectivity defines uniquely where the attached hydrogen(s) should be localized. However, ambiguity is inherent for the side chains of Asn, Gln, His, Ser, Thr, Tyr. Luckily, their hydrogen positions can in many cases be deduced by studying the hydrogen bonds they make.

Some programs like WHATIF and ... can examine automatically all possibilities and pick the most likely answer. However, it is the task of the modeler to take the ultimate resposibility and inspect the binding site of interest in great detail.

Reference:

Hooft RW, Sander C, Vriend G. (1996). Positioning hydrogen atoms by optimizing hydrogen-bond networks in protein structures. Proteins 26:363-376.