Re: [ccp4bb] refinement of ligands
What do you mean - running REFMAC directly on the output file? Are you sure you have the same space group given in the mtz file and the PDB file? If the MR has placed your structure in P32 say, nd the input mtz has SG P31, you need to change the mtz header to include the now-known SG. There are various ways given on the Reflection Utility task in the GUI. Eleanor On 30 Jul 2012, at 11:04, Damian Niegowski wrote: Dear all, I am currently trying to refine co-crystallized ligand structures at 2.8-3.4 Å resolution. After initial molecular replacement the density in the maps, both 2Fo-Fc and Fo-Fc looks good and the ligand seams to have bound. However after running Refmac directly on the output files the maps get much worse. I am using a clean pdb, without ligand or water for the Phaser and subsequent Refmac runs. Also when refining with the ligand in the B factors are a lot higher for the ligand then the surrounding residues, only when lowering the occupancy to 0.7-0.8 for the ligand the B factors look better. Is that acceptable to do at this resolution? R-free is in the 0.24-0.27 range. There is only a marginal change in R-free with the addition of ligand. TLS refinement seams to help alot for overall R values but does not improve the maps. I have also tried Phenix with simulated annealing, rigid body and reference structures. So the question is how best to proceed with refinement at the rather low resolution?? Regards, Damian Damian Niegowski Ph.D. Institute of Medical Biochemistry and Biophysics Karolinska Institutet Scheeles väg 2 171 77 STOCKHOLM e-mail: damian.niegow...@ki.se phone: 0046 8 524 876 33 fax: 0046 8 736 04 39
[ccp4bb] refinement of ligands
Dear all, I am currently trying to refine co-crystallized ligand structures at 2.8-3.4 Å resolution. After initial molecular replacement the density in the maps, both 2Fo-Fc and Fo-Fc looks good and the ligand seams to have bound. However after running Refmac directly on the output files the maps get much worse. I am using a clean pdb, without ligand or water for the Phaser and subsequent Refmac runs. Also when refining with the ligand in the B factors are a lot higher for the ligand then the surrounding residues, only when lowering the occupancy to 0.7-0.8 for the ligand the B factors look better. Is that acceptable to do at this resolution? R-free is in the 0.24-0.27 range. There is only a marginal change in R-free with the addition of ligand. TLS refinement seams to help alot for overall R values but does not improve the maps. I have also tried Phenix with simulated annealing, rigid body and reference structures. So the question is how best to proceed with refinement at the rather low resolution?? Regards, Damian Damian Niegowski Ph.D. Institute of Medical Biochemistry and Biophysics Karolinska Institutet Scheeles väg 2 171 77 STOCKHOLM e-mail: damian.niegow...@ki.se phone: 0046 8 524 876 33 fax: 0046 8 736 04 39
Re: [ccp4bb] refinement of ligands
Dear Damian, The initial map after molecular replacement and BEFORE atomic refinement is maybe not the highest quality map, but definitively the map with the least model bias. If you see good density for you inhibitors in those maps, they almost certainly will have bound. In that case I would fit the inhibitor in those very first maps and continue from there. The inhibitor density probably got worse during refmac refinement without inhibitor since refinement programs try to minimize the difference between Fobs and Fcalc. If an inhibitor is present in Fobs, but not in the model (Fcalc), the refinement program might try to tweak the model such that this difference (the inhibitor) gets flattened out. This is called model bias. Modern refinement programs try to counteract this effect by using maximum likelyhood or other statistical methods, but apparently at the low resolution you have refmac did not have enough data to succesfully remove this model bias. You may want to try buster for refinement, since this program may suffer less from model bias. Buster also allows you to do a group occupancy refinement of you inhibitor, since it may indeed by present at partial occupancy. An Rfree of 0.24 would be ok, 0.27 is a little high but still not really worrying. If your Rfactors go down a lot with TLS, this means that you have a lot of movement in your protein and that your maps are as good as they will get. Best, Herman -Original Message- From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Damian Niegowski Sent: Monday, July 30, 2012 12:04 PM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] refinement of ligands Dear all, I am currently trying to refine co-crystallized ligand structures at 2.8-3.4 Å resolution. After initial molecular replacement the density in the maps, both 2Fo-Fc and Fo-Fc looks good and the ligand seams to have bound. However after running Refmac directly on the output files the maps get much worse. I am using a clean pdb, without ligand or water for the Phaser and subsequent Refmac runs. Also when refining with the ligand in the B factors are a lot higher for the ligand then the surrounding residues, only when lowering the occupancy to 0.7-0.8 for the ligand the B factors look better. Is that acceptable to do at this resolution? R-free is in the 0.24-0.27 range. There is only a marginal change in R-free with the addition of ligand. TLS refinement seams to help alot for overall R values but does not improve the maps. I have also tried Phenix with simulated annealing, rigid body and reference structures. So the question is how best to proceed with refinement at the rather low resolution?? Regards, Damian Damian Niegowski Ph.D. Institute of Medical Biochemistry and Biophysics Karolinska Institutet Scheeles väg 2 171 77 STOCKHOLM e-mail: damian.niegow...@ki.se phone: 0046 8 524 876 33 fax: 0046 8 736 04 39
