Hi Peter, Thanks for your reply.

This data can be processed to P3 or P321 with the R-fac almost the same

After that, I used Bales to do the MR, two solutions were found in P32

(dimmer) and one in P32 21 (monomer)

Then, I refine the structure using phenix.refine in both P32 and P32 21 sets, but the R/Rfree stopped at 0.19/0.25 and never goes down.

However, if I treat the P32 data as a twin (with twin law h,-h-k,-l ), the R/Rfree can down to 0.12/0.18.

Do an R vs R analysis by loading up a model as well in xtriage to get a clue about the extend of the pseudo symmetry.

I do the R vs R analysis with P32 in xtriage, but the R value seems very close to zero. And the D_ncs is high (0.95) in resolution range 9.9841-4.2623, which may indicate an incorrect point group? The results are as follow. --------------------------------------------- Analysing possible twin law : h,-h-k,-l --------------------------------------------- mean |H| : 0.018 (0.50: untwinned; 0.0: 50% twinned) mean H^2 : 0.001 (0.33: untwinned; 0.0: 50% twinned) Estimation of twin fraction via mean |H|: 0.482 Estimation of twin fraction via cum. dist. of H: 0.487 Britton analyses Extrapolation performed on 0.47 < alpha < 0.495 Estimated twin fraction: 0.455 Correlation: 0.9777 R vs R statistic: R_abs_twin = <|I1-I2|>/<|I1+I2|> Lebedev, Vagin, Murshudov. Acta Cryst. (2006). D62, 83-95 R_abs_twin observed data : 0.022 R_abs_twin calculated data : 0.090 R_sq_twin = <(I1-I2)^2>/<(I1+I2)^2> R_sq_twin observed data : 0.001 R_sq_twin calculated data : 0.006 Perfoming correlation analyses The supplied calculated data are normalized and artificially twinned Subsequently a correlation with the observed data is computed. Results: Correlation : 0.832594802609 Estimated twin fraction : 0.49 Maximum Likelihood twin fraction determination Zwart, Read, Grosse-Kunstleve & Adams, to be published. The estimated twin fraction is equal to 0.478 Log[likelihood]: 78991.418 twin fraction: 0.443 D_ncs in resolution ranges: 9.9841 -- 4.2623 :: 0.950 4.2623 -- 3.4280 :: 0.694 3.4280 -- 3.0083 :: 0.655 3.0083 -- 2.7394 :: 0.885 2.7394 -- 2.5466 :: 0.853 2.5466 -- 2.3986 :: 0.824 2.3986 -- 2.2800 :: 0.817 The correlation of the calculated F^2 should be similar to the estimated values. Observed correlation between twin related, untwinned calculated F^2 in resolutiuon ranges, as well as ewstimates D_ncs^2 values: Bin d_max d_min CC_obs D_ncs^2 1) 9.9841 -- 4.2623 :: 0.981 0.902 2) 4.2623 -- 3.4280 :: 0.970 0.482 3) 3.4280 -- 3.0083 :: 0.958 0.429 4) 3.0083 -- 2.7394 :: 0.954 0.783 5) 2.7394 -- 2.5466 :: 0.964 0.728 6) 2.5466 -- 2.3986 :: 0.953 0.679 7) 2.3986 -- 2.2800 :: 0.963 0.667 Best Regards, Lauren