Hi Niu, Several things come to mind. First, it may not be trivial when the first component to be placed is ~20kDa and the second component (SU) is ~43kDa. The signal after placing the first component may be weak. Also, if the model for the smaller SU has low sequence identity with the target and you have 4 Angstrom data, it would not be surprising to encounters problems after searching with the smaller SU. But the question of why MR with MBP failed is worth investigating.
Here are couple of quick questions: (1) When you say you got some better results with the smaller component, what are the TFZ and LLG values? (2) Assuming you have very good 4A data, the correct space group, and a successful rotation and translation search with the first component so that maps are not entirely crappy, do you actually see a big mass of unaccounted density that could potentially fit MBP when you look at the maps after placing the first component? If yes, how about physically placing MBP in there yourself and starting from there. (3) What is the sequence identity of model with the target 20kDa component? (4) Do you know for sure that you have MBP in the crystal? Unlikely concern but I ask nonetheless. Also, I'm curious whether you ran Phaser jobs with the default settings or whether you tried tweaking some of the parameters? I'm happy to speak further about this offline. Good luck! Raji -- Raji Edayathumangalam Instructor in Neurology, Harvard Medical School Research Associate, Brigham and Women's Hospital Visiting Research Scholar, Brandeis University On Fri, May 16, 2014 at 11:03 AM, Niu Tou <niutou2...@gmail.com> wrote: > Dear All, > > Recently we collected some data of a MBP fusion protein, at around 4A > resolution. The protein itself is about half of the MBP size. However when > we tried to solve it with MR, it failed. We tried to use MBP alone, > homology model of target protein alone, and MBP+model. It is very strange > that MBP alone can not yield any reasonable solution at all, so does > searching with MBP and model together. While searching with model alone > could get some better results, but when fix it to search MBP, it failed. > There are 1 molecule per ASU with solvent content 55%. The spacegroup > should be right and we tried to search all possible alternatives in each > run, we also tried to lower it down, but did not work either. When running > Phenix.phaser, there is a warning at the beginning saying eLLG suggests > placing of ensembles will be very difficult. > > I wonder if anybody has encountered similar situation before. Any > suggestions will be greatly appreciated! > > Regards, > Niu >
