I’ve had success with MRSAD as Kay suggests. One option is to try putting everything into Auto-Rickshaw. It can run multiple cycles of MRSAD, auto-building a model for improved molecular replacement, finding new heavy atom sites and repeating.
There is a paper on this available: http://dx.doi.org/10.1107/S0907444909029643 And an Auto-Rickshaw server: http://www.embl-hamburg.de/Auto-Rickshaw/ Cheers, Jason. — Dr Jason Busby Laboratory of Structural Biology School of Biological Sciences The University of Auckland Private Bag 92019 Auckland Mail Centre Auckland New Zealand ph: +64 9 3737599 ext 88958 On 14/10/2014, at 6:48 am, Gabriel Salzman <[email protected]> wrote: > Hello, > > I am wondering if anyone has experience or advice for my current situation: > > I have a native dataset of a ~50kD protein complex at 2.4angstroms. One of > the members of the complex (~10kD) was able to be found easily by molecular > replacement. The space group was determined as P65. I was able to use > phenix.mr_rosetta to generate a partial model of the other member of the > complex (~half of this 40kD protein). When I do molecular replacement for > these two members of the complex in PHASER, the TFZ score is ~20-30. > Therefore, I am confident that this partial model is accurate. > > I recently soaked some complex crystals in a variety of heavy atoms to obtain > experimental phase information. My best signal is from an iodide soak. These > crystals diffracted to 3 angstroms and I can see a very weak but significant > anomalous signal to 5-4.5 angstroms. I have used autoSHARP to determine that > I have one site of ~99% occupancy and 4 other sites of <50% occupancy. I am > inclined to think these other sites are not true. My density modified map > seems to me to be better if I only include the single high-occupancy site in > the phase determination. Even with this weak experimental phase information, > my map is not good enough to see the substantial density from the piece of my > protein that was not included in the MR model. > > I have also tried phenix.phaser_EP with mediocre results. > > I apologize for the long question, and I would greatly appreciate any > suggestions for how to combine all of these data into a workable map so I can > build out the rest of the structure. > > Thanks, > Gabriel
