Chris, On Sun, Feb 23, 2014 at 2:52 PM, Chris Fage <cdf...@gmail.com> wrote:
> As suggested, I tried NCS and TLS in phenix.refine, although my R-factors > did not budge. > (...) > Forgive my ignorance, but I am not sure how to check whether the bulk > solvent model is reasonable. > I figure you used phenix.refine for refinement in which case bulk-solvent modeling and overall anisotropic scaling should be optimal within the framework of the implemented model used to describe it (for more info see http://journals.iucr.org/d/issues/2013/04/00/dz5273/dz5273.pdf). You can check it by looking in refinement log file (this is the table I always look first thing before doing anything else). There you will see something like this: Resolution Compl Nwork Nfree R_work <Fobs> <Fmodel> kiso kani kmask 43.996-14.23 96.15 88 12 0.1175 94.939 93.450 1.00 0.148 0.288 14.224-11.37 98.06 94 7 0.1211 84.547 83.471 1.00 0.151 0.295 11.321-9.092 98.90 156 23 0.0967 85.623 85.288 1.00 0.148 0.300 9.082-7.268 97.97 305 33 0.1400 61.020 59.892 1.00 0.143 0.300 7.264-5.810 98.96 601 63 0.1495 54.915 53.404 1.00 0.140 0.275 5.806-4.642 98.67 1134 127 0.1271 71.906 71.017 1.00 0.149 0.274 4.642-3.711 97.88 2198 203 0.1310 68.399 67.585 1.00 0.156 0.273 3.710-2.966 97.16 4129 462 0.1611 46.881 46.262 1.00 0.157 0.034 2.966-2.371 94.27 7737 874 0.1878 25.826 25.084 1.00 0.154 0.000 2.371-2.200 90.76 3826 418 0.1758 19.757 19.072 1.00 0.158 0.000 Pay attention to: - completeness (second column). Ideally in all resolution bins it should be greater than 90-95%. Smaller (especially at low resolution) completeness may result in corrupted maps. - kmask (last column) value in the lowest resolution bin: it should be around 0.2-0.5 (Acta Cryst. (2002). D58, 1387-1392). This characterize sanity of bulk-solvent model. - if a resolution bin has outstandingly high r-factor (5th column) this may indicate a problem. Pavel
