lower Rfree in phenix.maps compared to phenix.refine and phenix.reciprocal_space_arrays
Dear bb I have encountered a quite significant difference between Rfree reported by phenix.maps, refine and reciprocal_space_arrays. I have used default values in the GUI, and don't find differences in reflections statistics reported in the phenix.maps log that could explain the difference. I have another example also, but with a smaller difference between phenix.maps and phenix.refine. Both are quite large structures with maximum resolution between 3.5 and 4. Any explanation? phenix.maps ------------------------------------------------------------------------------- Bulk solvent correction and anisotropic scaling: |--(resolution: 3.68 - 29.45 A, n_refl.=22769 (all), 10.16 % free)------------| | | | r_work= 0.2550 r_free= 0.2823 coordinate error (max.-lik. estimate): 0.60 A | | | | x-ray target function (ml) for work reflections: 6.433959 | |-----------------------------------------------------------------------------| phenix.reciprocal_space_arrays |--(resolution: 3.68 - 29.45 A, n_refl.=22769 (all), 10.16 % free)------------| | | | r_work= 0.2741 r_free= 0.3045 coordinate error (max.-lik. estimate): 0.69 A | | | | x-ray target function (ml) for work reflections: 6.492710 | |-----------------------------------------------------------------------------| phenix.refine, strategy=none 1 macrocycle |--(resolution: 3.68 - 29.45 A, n_refl.=22769 (all), 10.16 % free)------------| | | | r_work= 0.2741 r_free= 0.3045 coordinate error (max.-lik. estimate): 0.69 A | | | | x-ray target function (ml) for work reflections: 6.492702 | |-----------------------------------------------------------------------------| gregers Professor Gregers Rom Andersen Department of Molecular Biology and Genetics Aarhus University Gustav Wiedsvej 10C DK8000 Aarhus C phone +45 871 55507 fax +45 861 23178 email [email protected] -------------extract of log files-------- phenix.maps log Parameters to compute maps:: maps { input { pdb_file_name = "/u/gra/myproject/juli2014/mave/ref_01.pdb" reflection_data { file_name = "/u/gra/myproject/juli2014/mave/reflections.mtz" labels = "FP,SIGFP" high_resolution = None low_resolution = None outliers_rejection = True french_wilson_scale = True french_wilson { max_bins = 60 min_bin_size = 40 } sigma_fobs_rejection_criterion = None sigma_iobs_rejection_criterion = None r_free_flags { file_name = "/u/gra/myproject/juli2014/mave/reflections.mtz" label = "R-free-flags43" test_flag_value = 1 ignore_r_free_flags = False } } } =============================== Reflection data =============================== F-obs: /u/gra/myproject/juli2014/mave/reflections.mtz:FP,SIGFP Miller array info: /u/gra/myproject/juli2014/mave/reflections.mtz:FP,SIGFP Observation type: xray.amplitude Type of data: double, size=22776 Type of sigmas: double, size=22776 Number of Miller indices: 22776 Anomalous flag: False Unit cell: (168.063, 154.031, 78.988, 90, 90, 90) Space group: P 21 21 2 (No. 18) Systematic absences: 0 Centric reflections: 2864 Resolution range: 29.4462 3.67631 Completeness in resolution range: 0.992029 Completeness with d_max=infinity: 0.989401 Wavelength: 0.0000 Number of F-obs in resolution range: 22776 Number of F-obs<0 (these reflections will be rejected): 0 Number of F-obs=0 (these reflections will be used in refinement): 0 Refinement resolution range: d_max = 29.4462 d_min = 3.6763 R-free flags: /u/gra/myproject/juli2014/mave/reflections.mtz:R-free-flags43 Miller array info: /u/gra/myproject/juli2014/mave/reflections.mtz:R-free-flags43 Observation type: None Type of data: int, size=22776 Type of sigmas: None Number of Miller indices: 22776 Anomalous flag: False Unit cell: (168.063, 154.031, 78.988, 90, 90, 90) Space group: P 21 21 2 (No. 18) Systematic absences: 0 Centric reflections: 2864 Resolution range: 29.4462 3.67631 Completeness in resolution range: 0.992029 Completeness with d_max=infinity: 0.989401 Wavelength: 0.0000 Test (R-free flags) flag value: 1 Number of work/free reflections by resolution: work free %free bin 1: 29.4471 - 7.8747 [2434/2441] 2190 244 10.0% bin 2: 7.8747 - 6.2702 [2347/2348] 2103 244 10.4% bin 3: 6.2702 - 5.4834 [2297/2300] 2076 221 9.6% bin 4: 5.4834 - 4.9847 [2284/2286] 2053 231 10.1% bin 5: 4.9847 - 4.6289 [2272/2274] 2045 227 10.0% bin 6: 4.6289 - 4.3569 [2294/2295] 2066 228 9.9% bin 7: 4.3569 - 4.1393 [2240/2241] 1994 246 11.0% bin 8: 4.1393 - 3.9595 [2254/2255] 2027 227 10.1% bin 9: 3.9595 - 3.8074 [2277/2278] 2055 222 9.7% bin 10: 3.8074 - 3.6763 [2077/2241] 1853 224 10.8% overall 20462 2314 10.2% ------------------------------------------------------------------------------- Input model file: /u/gra/myproject/juli2014/mave/ref_01.pdb ============================== Scattering factors ============================= ----------X-ray scattering dictionary---------- Number of scattering types: 6 Type Number sf(0) Gaussians S 48 15.91 1 P 28 14.91 1 Mg 2 11.95 2 O 2498 7.97 1 N 2146 6.97 1 C 7880 5.97 1 sf(0) = scattering factor at diffraction angle 0. Number of scatterers: 12602 At special positions: 0 Unit cell: (168.063, 154.031, 78.988, 90, 90, 90) Space group: P 21 21 2 (No. 18) ------------------------------------------------------------------------------- Bulk solvent correction and anisotropic scaling: |--(resolution: 3.68 - 29.45 A, n_refl.=22769 (all), 10.16 % free)------------| | | | r_work= 0.2550 r_free= 0.2823 coordinate error (max.-lik. estimate): 0.60 A | | | | x-ray target function (ml) for work reflections: 6.433959 | |-----------------------------------------------------------------------------| phenix.reciprocal_space_arrays Default params:: hkl_file = None pdb_file = None f_obs_label = None r_free_flags_label = None remove_f_obs_outliers = True bulk_solvent_and_scaling = True hendrickson_lattman_coefficients_label = None output_file_name = None space_group = None unit_cell = None job_title = None |--(resolution: 3.68 - 29.45 A, n_refl.=22769 (all), 10.16 % free)------------| | | | r_work= 0.2741 r_free= 0.3045 coordinate error (max.-lik. estimate): 0.69 A | | | | x-ray target function (ml) for work reflections: 6.492710 | |-----------------------------------------------------------------------------| |-----------------------------------------------------------------------------| | Bin Resolution Compl. No. Refl. R-factors Targets | |number range work test work test work test| | 1: 29.4471 - 9.3567 0.99 1316 144 0.1750 0.1935 7.111 7.2567| | 2: 9.3567 - 7.4666 1.00 1244 147 0.1844 0.2442 6.8649 7.0955| | 3: 7.4666 - 6.5345 1.00 1243 135 0.2403 0.3183 6.6514 6.7683| | 4: 6.5345 - 5.9424 1.00 1221 136 0.2622 0.3386 6.5345 6.547| | 5: 5.9424 - 5.5195 1.00 1230 131 0.2482 0.3091 6.3934 6.5161| | 6: 5.5195 - 5.1959 1.00 1199 141 0.2741 0.2699 6.3987 6.4315| | 7: 5.1959 - 4.9370 1.00 1196 129 0.2528 0.2600 6.3987 6.4955| | 8: 4.9370 - 4.7230 1.00 1227 136 0.2747 0.3240 6.4448 6.5001| | 9: 4.7230 - 4.5418 1.00 1178 151 0.2867 0.3278 6.4205 6.5232| | 10: 4.5418 - 4.3856 1.00 1224 113 0.2871 0.2980 6.4246 6.5162| | 11: 4.3856 - 4.2489 1.00 1201 151 0.3209 0.4007 6.4221 6.5583| | 12: 4.2489 - 4.1278 1.00 1174 137 0.3552 0.3395 6.4274 6.4033| | 13: 4.1278 - 4.0194 1.00 1190 129 0.3917 0.3944 6.4059 6.366| | 14: 4.0194 - 3.9216 1.00 1184 149 0.3931 0.3915 6.3718 6.38| | 15: 3.9216 - 3.8326 1.00 1221 116 0.3952 0.3998 6.3442 6.3031| | 16: 3.8326 - 3.7512 1.00 1168 138 0.4077 0.3949 6.334 6.3252| | 17: 3.7512 - 3.6763 0.88 1040 130 0.4322 0.4220 6.3237 6.2334| |-----------------------------------------------------------------------------| ------------------------------- phenix.refine refine { strategy = individual_sites individual_sites_real_space rigid_body \ individual_adp group_adp tls occupancies group_anomalous main { number_of_macro_cycles = 1 ============================= updating all scales ============================= start: r_work=0.3857 r_free=0.4073 n_reflections: 22776 start: r_work=0.3857 r_free=0.4073 (reset all scales to undefined) bulk-solvent and scaling: r_work=0.2758 r_free=0.3049 remove outliers: r_work=0.2758 r_free=0.3049 correct solvent mask: r_work=0.2741 r_free=0.3045 final: r_work=0.2741 r_free=0.3045 n_reflections: 22769 overall anisotropic scale matrix: V0: 0.0014,6.7700,1.3003,-6.1473,-1.1440,0.3484 V1: -0.0036,-0.0045,-0.0442,0.0324,0.0082,0.0028 ======================== Statistics in resolution bins ======================== Total model structure factor: F_model = k_total * (F_calc + k_mask * F_mask) k_total = k_isotropic * k_anisotropic Resolution Compl Nwork Nfree R_work <Fobs> <Fmodel> kiso kani kmask 29.446-20.724 97.09 90 10 0.2436 1579.386 1497.822 1.000 2.341 0.338 20.553-17.465 100.00 91 10 0.2757 1238.791 1140.148 1.000 1.886 0.335 17.444-14.852 99.35 137 16 0.2182 1221.185 1176.515 1.000 1.685 0.327 14.833-12.617 98.73 210 23 0.1958 1278.964 1244.386 1.000 1.494 0.307 12.608-10.717 98.98 349 39 0.1500 1428.238 1389.811 1.000 1.395 0.300 10.715-9.108 99.84 547 62 0.1416 1410.502 1384.438 1.000 1.354 0.290 9.104-7.738 100.00 883 99 0.1789 1027.723 1002.711 1.000 1.233 0.290 7.736-6.574 99.94 1427 159 0.2385 682.594 646.245 1.000 1.076 0.290 6.573-5.586 99.88 2288 254 0.2567 515.856 486.005 1.000 0.892 0.270 5.586-4.747 99.90 3710 411 0.2654 489.477 455.553 1.000 0.818 0.220 4.746-4.033 99.95 5973 671 0.3225 441.426 390.526 1.000 0.817 0.170 4.033-3.676 96.99 4751 559 0.4048 366.235 296.284 1.000 0.848 0.170 Approximation of k_total with k_overall*exp(-b_overall*s**2/4) k_overall=1.1939 b_overall=26.4160
Hi Gregers, this is confusing I agree, thanks for pointing this out! phenix.maps applies a correction to bulk-solvent mask to not fill bulk-solvent into internal "bulk-solvent free" voids in protein; it reports R-factors after this correction is done. Other tools also do this correction for the purpose of map calculation but they do not report corresponding R-factors (after such correction). I will review this to make reported output clearer. Pavel On 7/31/14, 12:08 AM, Gregers Rom Andersen wrote:
Dear bb
I have encountered a quite significant difference between Rfree reported by phenix.maps, refine and reciprocal_space_arrays. I have used default values in the GUI, and don't find differences in reflections statistics reported in the phenix.maps log that could explain the difference. I have another example also, but with a smaller difference between phenix.maps and phenix.refine. Both are quite large structures with maximum resolution between 3.5 and 4. Any explanation?
phenix.maps
-------------------------------------------------------------------------------
Bulk solvent correction and anisotropic scaling:
|--(resolution: 3.68 - 29.45 A, n_refl.=22769 (all), 10.16 % free)------------|
| |
| r_work= 0.2550 r_free= 0.2823 coordinate error (max.-lik. estimate): 0.60 A |
| |
| x-ray target function (ml) for work reflections: 6.433959 |
|-----------------------------------------------------------------------------|
phenix.reciprocal_space_arrays
|--(resolution: 3.68 - 29.45 A, n_refl.=22769 (all), 10.16 % free)------------|
| |
| r_work= 0.2741 r_free= 0.3045 coordinate error (max.-lik. estimate): 0.69 A |
| |
| x-ray target function (ml) for work reflections: 6.492710 |
|-----------------------------------------------------------------------------|
phenix.refine, strategy=none 1 macrocycle
|--(resolution: 3.68 - 29.45 A, n_refl.=22769 (all), 10.16 % free)------------|
| |
| r_work= 0.2741 r_free= 0.3045 coordinate error (max.-lik. estimate): 0.69 A |
| |
| x-ray target function (ml) for work reflections: 6.492702 |
|-----------------------------------------------------------------------------|
gregers
Professor Gregers Rom Andersen
Department of Molecular Biology and Genetics
Aarhus University
Gustav Wiedsvej 10C DK8000 Aarhus C
phone +45 871 55507 fax +45 861 23178
email [email protected]
participants (2)
-
Gregers Rom Andersen
-
Pavel Afonine