I believe the literature consensus is not to detwin until R values get below ~40%, and if you're still doing DM, I assume the R's are still pretty poor. If you've placed a model with MR, you could just put it into Refmac with detwinning.
Also, I had pretty good success in a twinned structure recently with the following procedure: --Used partial model phases in Phaser to find weak anomalous scatterers, phased from them, then autobuilt. --Refined this model in Refmac using twin refinement --Manual rebuild, Refmac, then back to square one. I think this worked well for removing the notoriously bad model bias in twinned structures, since the anomalous information is, I believe, orthogonal to the real scattering (pun intended). But I mean "orthogonal" in the sense of "independent." (Please let me know if you think that statement is false--I think it depends on the details of the Phaser calculation, but I don't think it uses prior knowledge of scatterers (met/cys/ions) from the input model to calculate anomalous llg maps). Jacob -----Original Message----- From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Birtley, James Sent: Friday, March 06, 2015 12:52 PM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] Detwinned Fobs for density modification Dear CCP4 community, Does anyone know if the F-obs-filtered array in the map file output by phenix.refine from a twin refinement has detwinned Fs? Or if not how to obtain these? We’re trying to set up a density modification run using data from a P41 dataset with twin fraction ~0.5 and two mol / au. Thanks a lot. James
