I have had a similar experience.
I was trying to model a bunch of waters to simulate an unknown ligand (UNL) in an unmodeled density. The waters were all in different alternate conformations of the same residue number. When the occupancies of these waters were set to 0.5, the vdw repulsion became absent; waters stayed within density at close distances. But, as soon as the occupancies were changed to anything but 0.5, waters got pushed out of density due to vdw repulsion. So my feeling is that at 0.5 occupancy the atom does not see vdw repulsion (?). Change it to something else and atom should crawl back into its density. I was using refmac5 version 5.2.0019. Thanks Abhinav Stanford Synchrotron Radiation Laboratory Joint Center for Structural Genomics Mail Stop 99 Phone: (650) 926-2992 Fax: (650) 926-3292 ________________________________ From: CCP4 bulletin board [mailto:[EMAIL PROTECTED] On Behalf Of Garib Murshudov Sent: Wednesday, July 30, 2008 5:18 AM To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] Preventing close contact between protein and ligand Dear Snageetha 1) Could you check please if specified atoms have zero occupancy. Atoms with zero occupancy are considered as absent and there are not restraints on them 2) symm y at the end of instructions means that the program check all possible symmetry operators and finds minimal distance. Most probably 5.024 is the distance between symmetry related atoms 3) to remove antibumping between different chains there is an undocumented keyword. It can be used. the keyword is (as an example) vdwrestraints exclude between chains A B Please let me know if this instruction does not work. NB: This option should not be used unless you know what you are doing (that is the reason why it has not been documented). If there are clashes between chains then there are reasons for that. For example if ligand has half occupancy then it is very likely that surrounding atoms also have multiple conformation and you should model them. regards Garib On 29 Jul 2008, at 18:24, Sangeetha Vedula wrote: Dear bb users, I am refining a protein-ligand complex (at 1.68 A resolution) in which the ligand lies on a 2-fold crystallographic symmetry axis. The ligand occupancy is, therefore, 0.5 in each asymmetric unit. I am almost at the end of the refinement but one problem has me stumped. Refmac keeps moving a carbon in the ligand too close to a serine OG and an oxygen too close to an arginine CD. Given that the ligand is at the interface, the density is not perfect. However, I rebuild the ligand to eliminate close contacts and still be within density and refmac pulls it right back close to the protein. The refined position does not even look better than the rebuilt one! It almost always looks worse! Would refmac put less weight on close contacts with the ligand because it is only partially occupied? I tried to use external restraints between the ligand and the residues so that they are kept further away. Upon searching the net, I found this command line: external distance first chain [ch] residue [res] insertion [ins] - atom [n] [altcode [a]] second chain [ch] residue [res] insertion [ins]- atom [n] [altcode [a] ] value [v] sigma [s] [symm y/n] I thought (hoped) that the distance herein is the minimum distance of approach between the specified atoms, I added these lines from within "Developer options" in refmac interface: exte dist first chain A resi 59 atom CD seco chain X resi 2001 atom O1 valu 3.2 sigm 0.02 symm Y exte dist first chain A resi 27 atom OG seco chain X resi 2001 atom C10 valu 3.2 sigm 0.02 symm Y It didn't recognize these restraints at all. However, when I change these lines to: exte dist first chain A resi 59 atom CA seco chain X resi 2001 atom O1 valu 3.2 sigm 0.02 symm Y exte dist first chain A resi 27 atom OG seco chain X resi 2001 atom C10 valu 3.2 sigm 0.02 symm Y Refmac recognizes the first line but not the second - lines from log file: Bond distance deviations from the ideal >10.000Sigma will be monitored A 59 ARG CA . - X 2001 DIE O1 . mod.= 5.024 id.= 3.200 dev= -1.824 sig.= 0.020 This raises two concerns: Concern 1: From the first line of output: the restraints here don't seem to be minimizing close contact at all; it seems to think they are bonded somehow (the distance between these atoms is not 5.024; it is 6.26 A; I don't know what 5.024 A is!). I am missing something here. It'd be great if someone can tell me what that is! Concern 2: This command only works when the first atom specified is a C-alpha atom (or maybe a main chain atom; I didn't try using other main chain atoms). Why is that? AND ULTIMATELY, is there some way I can tell refmac not to make the ligand and protein clash? I'd really appreciate any help! Thanks, Sangeetha.