Is the data anisotropic? This would explain the strange processing statistics beyond 3.0 Angstrom. If using Dials directly it is worth looking for abnormalities with the reciprocal lattice viewer.
I've seen similar issues with the completeness being low, when data is weak in one direction, despite reflections being reported in the scaling output. It may also be worth running some analysis of the data with Staraniso (https://staraniso.globalphasing.org/cgi-bin/staraniso.cgi) to try to help. Andy Quoting Martin Malý <[email protected]>:
Dear Matt, thank you for asking. I am bit confused how you set up the PAIREF run. In paired refinement, data are added step by step - going to higher resolution, i.e. lower d_hkl. So the merged data in the PAIREF input should not be cut to much. The data statistics look from DIALS quite strange, there is very good completeness and CC1/2 for high resolution shells, but very poor <I/sigma> and Rmeas. I would suggest to refine the model at 3 A to be safe. Then, run PAIREF with the initial resolution 3 A and using the cutoffs 2.8, 2.6, 2.4 and 2.2 A. Diffraction data must be up to 2.2 A in that case, no truncation to 2.4 A as you mentioned previously. I hope it will help. Please feel free to ask and let me know if it was successful. Best regards, Martin On 30. 09. 22 17:29, Matt McLeod wrote:Hi all, I am having a bit of trouble with PAIREF. I am trying to determine the resolution cutoff of a dataset which has a high Rwork (0.223)/free (0.275)value for the resolution (2.24A). I have truncated this data further to 2.4A and the R values get better but the electron density maps do not improve whatsoever. When I put this data into PAIREF I get this message. WARNING: There are only 40 < 50 free reflections in the resolution shell 2.75-2.70 A. Values of statistics Rfree and CCfree in this shell could be misleading. Consider setting thicker resolution shells. This error occurs for all resolution bins. Looking at the pdb log of the refined structure (where the data looks quite reasonable, maybe not a 2.3 dataset but not much worse). REMARK 3 FIT TO DATA USED IN REFINEMENT (IN BINS). REMARK 3 BIN RESOLUTION RANGE COMPL. NWORK NFREE RWORK RFREE CCWORK CCFREE REMARK 3 1 60.08 - 4.28 1.00 4135 213 0.1646 0.2213 0.940 0.901 REMARK 3 2 4.28 - 3.40 1.00 4095 201 0.1951 0.2672 0.928 0.857 REMARK 3 3 3.40 - 2.97 1.00 4099 204 0.2736 0.3303 0.861 0.739 REMARK 3 4 2.97 - 2.70 1.00 4026 212 0.3486 0.3508 0.726 0.551 REMARK 3 5 2.70 - 2.51 0.78 3178 154 0.4027 0.4062 0.637 0.594 REMARK 3 6 2.50 - 2.36 0.29 1154 60 0.4118 0.3984 0.647 0.547 REMARK 3 7 2.36 - 2.24 0.01 29 1 0.3366 0.4089 0.343 0.000 Where I assume NFree is the issue, but doesn't suggest that there arent enough reflections. There needs to be a higher cutoff as the completeness goes to pot, but the scaling log from DIALS suggested it was at least a good starting point. Statistics by resolution bin: d_max d_min #obs #uniq mult. %comp <I> <I/sI> r_mrg r_meas r_pim cc1/2 cc_ano 118.81 5.97 10436 1577 6.62 99.94 797.6 22.8 0.110 0.120 0.046 0.992* -0.397 5.97 4.74 9602 1549 6.20 100.00 491.7 10.0 0.172 0.188 0.075 0.979* -0.240 4.74 4.14 10410 1528 6.81 100.00 513.5 10.3 0.191 0.207 0.079 0.978* -0.281 4.14 3.76 10833 1551 6.98 100.00 324.7 6.9 0.238 0.257 0.098 0.969* -0.419 3.76 3.49 10956 1526 7.18 100.00 207.5 4.6 0.259 0.280 0.105 0.975* -0.271 3.49 3.29 10899 1546 7.05 100.00 133.4 3.1 0.272 0.294 0.111 0.966* -0.258 3.29 3.12 9565 1528 6.26 99.93 86.9 2.0 0.314 0.343 0.136 0.935* -0.346 3.12 2.99 9256 1519 6.09 100.00 57.1 1.3 0.354 0.387 0.154 0.932* -0.354 2.99 2.87 9317 1524 6.11 100.00 43.6 1.0 0.403 0.441 0.177 0.934* -0.198 2.87 2.77 10087 1549 6.51 100.00 30.9 0.7 0.455 0.494 0.192 0.923* -0.116 2.77 2.69 10157 1524 6.66 100.00 26.4 0.6 0.503 0.546 0.210 0.927* -0.132 2.69 2.61 10351 1541 6.72 100.00 19.3 0.4 0.613 0.665 0.255 0.925* -0.155 2.61 2.54 10121 1504 6.73 100.00 15.5 0.4 0.735 0.797 0.306 0.906* -0.088 2.54 2.48 10510 1539 6.83 100.00 12.4 0.3 0.932 1.009 0.385 0.857* 0.006 2.48 2.42 10432 1519 6.87 100.00 11.2 0.2 0.972 1.052 0.399 0.829* -0.111 2.42 2.37 10596 1532 6.92 100.00 10.2 0.2 1.095 1.184 0.447 0.844* -0.083 2.37 2.32 10577 1535 6.89 100.00 8.9 0.2 1.259 1.361 0.514 0.825* -0.082 2.32 2.28 10217 1513 6.75 100.00 7.8 0.2 1.421 1.540 0.587 0.725* -0.056 2.28 2.24 9702 1521 6.38 100.00 6.4 0.1 1.683 1.834 0.721 0.600* -0.001 2.24 2.20 9551 1538 6.21 100.00 6.1 0.1 1.895 2.071 0.826 0.603* -0.041 118.64 2.20 203575 30663 6.64 99.99 142.0 3.3 0.222 0.241 0.093 0.989* -0.265 Does anyone have any insight into where PAIREF is getting hung up? I am welcome to any other suggestions with regards to handling this dataset to determine resolution cutoff. I can also supply any other log files that may be useful Matt ######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1 This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1 This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1 This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
