Hi Joe, yes, I agree on this. It's on my list already and hopefully I do it soon. Thanks for the feedback! Pavel. On 9/30/09 4:41 PM, Joe Krahn wrote:
A few more comments on writing peak coordinates, after going through peaks based on the current PHENIX log file:
Having actual coordinates present as atoms makes it easy to measure distances to other atoms. Less importantly, if there are two peaks close to each other, it is not always obvious which peak matches the log.
It may be sufficient to write the peak coordinates in the log file. That would avoid the need for adding another output file. It would not be that hard to parse into either a PDB file, or some sort of program-specific list or script.
Thanks, Joe Krahn
Krahn, Joe (NIH/NIEHS) [C] wrote:
Pavel Afonine wrote:
Hi Joe,
PHENIX has an option to find positive and negative peaks, but only lists them in the log file along with the residue they are closest to.
I have done it this way having the following in mind:
- you load you refined model and maps in your favorite graphics program, Coot for example; - you take that +/- peaks list; - and you go residue-by-residue and check the flagged peaks.
Is it possible to write out the actual peak coordinates, such as to a PDB file?
It's possible of course, but I have a couple of questions:
- why to do this? - what do I put as atom name, residue name, occupancy, B-factor, atom chemical type ?
Thanks again for your feedback! Pavel.
Most of the time, I optimize a model by stepping through peaks, rather than walking through all residues. In most software, it is easy to step through residues in a PDB file, whereas converting the PHENIX log output into something that can automate this is likely program specific.
My standard approach is to get a peak list in the form of a PDB file, sorted from the highest peak, and work through those peaks until they become too weak. I can fix model errors, or add a water at the peak position, or decide that it is something other than water. Having an atom at the peak makes it easy to add waters, which is good for lower-resolution or noisy maps, where I don't trust automated waters.
Many programs write peaks as PDB files, usually with B-factor set from the difference-map sigma level. Some programs write them out as waters, which is probably safest to avoid problems with non-standard names. The occupancy is usually 1.0, but it could be set to zero to flag that they are not actual model coordinates.
This approach used to be very common. Some new programs have built-in peak search utilities, so it may be getting less common for that reason.
Thanks, Joe Krahn
_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb