HTP ligand screening

From: Jacob Keller
Date: 26 August 2011 21:43

Dear Crystallographers,

I have ~30 data sets from ligand soaks of my protein of known
structure (all approximately the same cell. Can anyone suggest a high
throughput method which would do molecular replacement, refine, then
output new blobs, perhaps as a water pdb file? I am sure they do this
or better in pharma every day...

Jacob Keller

*******************************************
Jacob Pearson Keller
Northwestern University
Medical Scientist Training Program
*******************************************

----------
From: Bosch, Juergen

See message to Greg :-)

CAD them first, run MR with one and then just refine using different labels, inspect your difference density maps and enjoy your success rate of 6% :-)

You can certainly write a tiny script which would use different labels and do a quick phenix.refine or Rafmac.

I would skip the water picking as you can't skip the part where you actually look at the density maps. So I'd rather do a few cycles of refinement and visually inspect using Coot find unmodelled blobs and go from there. For 30 data sets that shouldn't take much more than a day and at the end you know which ones look promising and focus on those first.

Jürgen

......................

Jürgen Bosch

Johns Hopkins University

Bloomberg School of Public Health

Department of Biochemistry & Molecular Biology

Johns Hopkins Malaria Research Institute

615 North Wolfe Street, W8708

Baltimore, MD 21205

----------
From: <Alexander.Schiffer

Dear Jacob,

Globalphasing has a new programm called Pipedream that should do just
what you want. I am not sure if it is available to non consortium
members.

Best regards,

Alexander
-------------


From: John Badger 
Date: 30 August 2011 03:36

Pipelined data to map calculations can be run with the 'bng' script in MIExpert, a companion package to the MIFit software. These are open source packages at

http://code.google.com/p/mifit

The bng script can either be run on the command-line or through the MIFit GUI. Both GUI and processing script are written in python and fully exposed so it you don't like the defaults or one of the underlying CCP4 processing engines gets updated with new keywords or some other change you can easily fix.

Possibly novel features are the ability to start from image data (assuming it is 'reasonably good') and the ability to load up a whole series of data sets in a single calculation. i.e. you can setup all 30 to run and then go away for a few hours before reviewing the refined maps.

Outputs include various processing summaries, refined phased data and coordinates before and after water picking. If you happen to have an automated ligand-fitting technology you can throw that in also.

I use this script all the time for projects (usually cocrystals or fragment screening) where I have many data sets and I want an efficient route to the first maps.

Thanks
John Badger