Re: [ccp4bb] R values are higher in SF-CHECK

2010-07-06 Thread Pavel Afonine
Hi Rakesh, providing everything else is good about your PDB file (for example, no post-refinement manipulations done to the model), phenix.model_vs_data should reproduce the R-factors, since it accounts for TLS (both, PHENIX or REFMAC formatted), it does the advanced bulk-solvent and anisotro

Re: [ccp4bb] R values are higher in SF-CHECK

2010-07-06 Thread Dirk Kostrewa
Hello Rakesh, yes, you are right. In contrast to all modern refinement programs, sfcheck neither applies TLS nor the superior mask bulk solvent correction (it applies the Babinet bulk solvent correction by double-exponential scaling). As a result, the R-factors from sfcheck are higher than t

[ccp4bb] R values are higher in SF-CHECK

2010-07-06 Thread Rakesh Joshi
Hello, I refined my structures using REFMAC5 ( using TLS in combination ). When i used the SF-Check program, my R-values are ~5% higher than in those reported by refmac. Is this due to the fact that SF-check does bulk solvent calculations differently and does not account for TLS refinement? A