From: Pawel
Date: 3 September 2011 20:07
Dear all,
First of all just "hello" to everyone since I am new to this forum. Not being a crystallographer, bear with me if I say something crazy.
I was recently tinkering with Sfall and noticed that the Fcalc output by Sfall have different values depending on if one feeds Sfall with an input mtz file containing Fobs ("mode sfcalc xyzin" vs. "mode sfcalc xyzin hklin") and also depending on if one tells Sfall to scale the Fobs or not ("noscale" on or off). This puzzles me as I would have thought that since Fcalc are calculated from the atomic coordinates, they should be affected by neither the Fobs fed to Sfall nor the scaling of the Fobs.
As an example, Sfall calculations on PDB:1AHO. In each case the Fcalc column is the penultimate one. Here the difference in Fcalc between scaling or not scaling the Fobs is not so pronounced, but it still puzzles me that there would be any change at all. Also, in this case when the Sfall is run without hklin, there is a substantial change in PHIC.
1. HKLIN was provided and scaling turned on.
LIST OF REFLECTIONS
===================
H K L FP SIGFP FC PHIC
0 1 2 162.32 3.86 238.41 270.00
0 1 4 97.72 2.34 190.32 90.00
0 1 5 230.11 12.78 256.90 270.00
0 1 6 205.83 7.00 322.93 90.00
0 2 2 53.28 2.14 198.16 0.00
0 2 4 137.66 5.46 2.63 0.00
0 2 5 249.32 5.84 293.85 0.00
0 2 6 203.96 5.37 303.16 180.00
2. HKLIN was provided but scaling was turned off.
LIST OF REFLECTIONS
===================
H K L FP SIGFP FC PHIC
0 1 2 158.53 3.77 238.28 270.00
0 1 4 95.44 2.29 189.92 90.00
0 1 5 224.74 12.48 256.07 270.00
0 1 6 201.02 6.84 321.43 90.00
0 2 2 52.04 2.09 198.01 0.00
0 2 4 134.45 5.33 2.63 0.00
0 2 5 243.50 5.70 292.83 0.00
0 2 6 199.20 5.24 301.69 180.00
3. HKLIN with Fobs was not provided (both FC and PHIC are quite different)
LIST OF REFLECTIONS
===================
H K L FC PHIC
0 1 2 260.16 -108.18
0 1 4 122.43 -7.47
0 1 5 38.13 81.93
0 1 6 209.85 125.96
0 2 2 386.47 -78.65
0 2 4 145.47 -82.02
0 2 5 107.70 -15.80
0 2 6 101.94 138.75
If anyone could explain, I'd be very grateful. Thanks!
Pawel Janowski
----------
From: Eleanor Dodson
Hmm - Firstly: your final FC PHIC calculation are just wrong. I cant tell why without more details - is the SG set correctly, because those phases cannot reflect the SG symmetry.. Can you send the log file?
For the scaling against Fobs Qs. It looks to me as though you are using values of Fobs already scaled to match the model by some program.. The scaling algorithms in different programs are subtly different - there is no clear "correct" answer in to how to handle a dolvent continuum for instance - so I suspect the differences reflect that.
In sfall (and I think most programs) the model B factors are changed to give best agreement with the Wilson scale for the observed data, and a scale factor is applied to the Fobs to bring it to match the fcalc.
So I guess sfall has changed the overall B for the model slightly changing the higher resolution Fcs a bit..
Eleanor
Date: 3 September 2011 20:07
Dear all,
First of all just "hello" to everyone since I am new to this forum. Not being a crystallographer, bear with me if I say something crazy.
I was recently tinkering with Sfall and noticed that the Fcalc output by Sfall have different values depending on if one feeds Sfall with an input mtz file containing Fobs ("mode sfcalc xyzin" vs. "mode sfcalc xyzin hklin") and also depending on if one tells Sfall to scale the Fobs or not ("noscale" on or off). This puzzles me as I would have thought that since Fcalc are calculated from the atomic coordinates, they should be affected by neither the Fobs fed to Sfall nor the scaling of the Fobs.
As an example, Sfall calculations on PDB:1AHO. In each case the Fcalc column is the penultimate one. Here the difference in Fcalc between scaling or not scaling the Fobs is not so pronounced, but it still puzzles me that there would be any change at all. Also, in this case when the Sfall is run without hklin, there is a substantial change in PHIC.
1. HKLIN was provided and scaling turned on.
LIST OF REFLECTIONS
===================
H K L FP SIGFP FC PHIC
0 1 2 162.32 3.86 238.41 270.00
0 1 4 97.72 2.34 190.32 90.00
0 1 5 230.11 12.78 256.90 270.00
0 1 6 205.83 7.00 322.93 90.00
0 2 2 53.28 2.14 198.16 0.00
0 2 4 137.66 5.46 2.63 0.00
0 2 5 249.32 5.84 293.85 0.00
0 2 6 203.96 5.37 303.16 180.00
2. HKLIN was provided but scaling was turned off.
LIST OF REFLECTIONS
===================
H K L FP SIGFP FC PHIC
0 1 2 158.53 3.77 238.28 270.00
0 1 4 95.44 2.29 189.92 90.00
0 1 5 224.74 12.48 256.07 270.00
0 1 6 201.02 6.84 321.43 90.00
0 2 2 52.04 2.09 198.01 0.00
0 2 4 134.45 5.33 2.63 0.00
0 2 5 243.50 5.70 292.83 0.00
0 2 6 199.20 5.24 301.69 180.00
3. HKLIN with Fobs was not provided (both FC and PHIC are quite different)
LIST OF REFLECTIONS
===================
H K L FC PHIC
0 1 2 260.16 -108.18
0 1 4 122.43 -7.47
0 1 5 38.13 81.93
0 1 6 209.85 125.96
0 2 2 386.47 -78.65
0 2 4 145.47 -82.02
0 2 5 107.70 -15.80
0 2 6 101.94 138.75
If anyone could explain, I'd be very grateful. Thanks!
Pawel Janowski
----------
From: Eleanor Dodson
Hmm - Firstly: your final FC PHIC calculation are just wrong. I cant tell why without more details - is the SG set correctly, because those phases cannot reflect the SG symmetry.. Can you send the log file?
For the scaling against Fobs Qs. It looks to me as though you are using values of Fobs already scaled to match the model by some program.. The scaling algorithms in different programs are subtly different - there is no clear "correct" answer in to how to handle a dolvent continuum for instance - so I suspect the differences reflect that.
In sfall (and I think most programs) the model B factors are changed to give best agreement with the Wilson scale for the observed data, and a scale factor is applied to the Fobs to bring it to match the fcalc.
So I guess sfall has changed the overall B for the model slightly changing the higher resolution Fcs a bit..
Eleanor
No comments:
Post a Comment