As others point out, anisotropy is endemic in protein crystals, and severe cases make the single number score for Rmerge, CC1/2 etc pretty meaningless for the outer shell. But do not throw out the meaningful data at your higher resolution because of single number stats- first LOOK at your images to get a feel for where usable spots disappear along your best axis.
Then refine using the usable data - do not cut on the basis of a single number. The higher reslution terms still help refinement, even if they only exist along one cell axis. And remember maximum likelihood weighting means the noise data along the poor axes are weighted very low, so although the R may seem higher than you like, the map quality may not be too bad. Eleanor On 18 June 2017 at 15:36, Alexandre Ourjoumtsev < alexander.ourjoumt...@univ-lorraine.fr> wrote: > Dear Khoa Pham, > > the question of a formal definition of en "effective" resolution and its > variability with space direction is discussed in > > a) Urzhumtseva, L., Klaholz, B.P., Urzhumtsev, A. (2013) "On effective > and optical resolution of diffraction data sets". *Acta Cryst., **D69**,* > 1921-1934. > b) Urzhumtseva, L., Urzhumtsev, A. (2015) "A program to calculate > effective resolution". *J. Appl.**Cryst., **48*, 589-597. > > U > With best regatrds, > > Sacha Urzhumtsev > > ----- Le 18 Juin 17, à 15:42, Khoa Pham <khoa.p...@einstein.yu.edu> a > écrit : > > Dear Gerard, > > Thank you so much for the very interesting discussions and the > recommendation of using STARANISO. I will submit my data to STARANISO > server and keep you updated. > > Sincerely, > > Khoa Pham > >