Considerations when using a coordinate file for modeling
I attended the Mid-Atlantic Graduate Student Symposium (MAGSS) in Medicinal Chemistry this week.
My goal was to see how structures from the Protein Data Bank (PDB) are used by the non-crystallographic community.
I had a brief discussion with a colleague who was doing a number of modeling experiments. She was, essentially, modeling how the protein would move over the course of 35 nanoseconds.
I asked her how she determines which model to use from the PDB. The two criteria were:
1) the PDB entry had been used in other publications
2) the entry had the highest resolution
The most useful piece of information in deciding how to determine the proper position of the coordinates is the electron density map. However, I believe there is currently no resource that displays the final electron density map generated by the crystallographer.
We are currently making a huge effort to generate and store the coordinates of proteins, but are we doing enough to make them useful?
What criteria should the modeling community be using to judge the quality of coordinates from the PDB?
Paul
June 25th, 2009 at 2:17 PM #
I personally think non-crystallographers should use the Electron Density Server at Uppsala as a first port of call to decide on the reliability of a structure. The real-space parameters can instantly flag particularly bad models, and there are easy to use java applets that allow you to look at ED maps without installing additional software (assuming the reflection data has been deposited).
Sean
July 1st, 2009 at 4:50 PM #
I think you are right on and feel that if this is combined with a simple to read algorithm, should steer people in the right direction.
http://www.p212121.com/2009/05/31/pdb-structure-quality-tools/