Dear Khoa, Thank you for your reply and for the news that your data do show strong anisotropy.
In such cases, the reflex is indeed to cut the data down to a resolution where the statistics look more reasonable. However, as these statistics are computed in spherical shells, i.e. continue to include data in the weak directions, getting those reasonable values necessarily entails rejecting significant data in the best directions. The analysis of anisotropy done by the STARANISO server I pointed you to will avoid this rejection: it will instead discard all the pure-noise data so that your model will be refined (and hence the Fc compared to Fo to calculate the statistics) only for data that are significantly above noise. As you are a newcomer, you might find the following introductory material useful: http://staraniso.globalphasing.org/anisotropy_about.html I therefore strongly advise you to submit your data to the server and to keep in touch off-list for any extra advice you might need. With best wishes, Gerard. -- On Sun, Jun 18, 2017 at 02:29:06AM +0000, Khoa Pham wrote: > Dear Gerard and James, > > Thank you very much for the suggestions. My data was collected at APS. > Yesterday, an expert in my institution kindly inspected my raw data and > pointed out to me that my data are indeed severely anisotropic in the b > direction. He also recommended me to cut the data to 3.27 A (or lower). I > took his advise and tried refining the data with a resolution of 3.27 A. I > found that it indeed reduced the values of Rwork/Rfree too much more > reasonable window (0.264/0.322, respectively). I will try improving the > quality of my crystals and hope to collect better data sets. > > I am a new comer to protein crystallography. I appreciate all for giving me > valuable advises. > > Sincerely, > > Khoa Pham