Re: [phenixbb] XN-refinement
this is good to hear, and this demonstrates the power of using more data. The only issue here, as I mentioned before, is that in joint XN refinement we refine one single model against two data sets, while ideally it should be two models in order to account for genuine unique information that is contained in X-ray and neutron data. This will be addressed at some point.
Out of curiosity, could you elaborate on why it would be preferred to refine two separate models? It seems to me, admittedly not having any experience with neutron diffraction, that the ultimate goal is to model the system of the crystal, and given 2 data sets, we would want to create the single model describing the structure in the crystal that best agrees with all the available data. Are the neutron and X-ray data so different that using them for joint refinement is disadvantageous?
I think I wrote that already in previous email.. Structures corresponding to two data sets are of course overall similar may be different in local details: - Different X-H(D) distances (see figure 4 here: http://journals.iucr.org/d/issues/2010/11/00/dz5209/index.html) - Different data collection temperatures: -- Different B factors; -- Different visible water structure; -- Different local disorder (that may be captured as distinct conformations at low temperature, but may be blurred at higher temperature); - Different completeness and resolution of one data set may allow features in one model corresponding to this data set that may not be supported by the other data set. For example, you may see a well resolved water with one or two D in neutron map but you may not even see any water at all in X-ray map, or the other way around. - etc etc etc. So it is advantageous to use both X-ray and neutron data sets together but these nuances need to be carefully taken care of which isn't the case in current implementations of joint XN refinement protocols. This makes a good room for future improvements. Pavel
Hello Pavel,
Collecting the X-ray and the neutron data at the same temperature leads to
the following, and quoting from your 2010 dz5209 article, which I agree
with:-
"If both data sets are collected from the same crystal (similar unit-cell
parameters and identical space group) under the same conditions
(temperature) then the structures derived from both data sets will be very
similar. This makes it possible to use both data sets in refinement
simultaneously [joint X-ray and neutron (XN) refinement]."
This is also important because atom labels are often amended by the PDB to
follow its internal 'rules' and can make direct comparison between two
structures awkward. This is particularly so when comparing structures at
two or more temperatures where atoms can naturally disappear at higher
temperatures ie then the one to one correspondence between the two or more
PDB file's atom labels is lost altogether.
Overall then to strive to follow your maxim above is a very good plan!
Greetings,
John
On Sat, Oct 15, 2016 at 4:40 AM, Pavel Afonine
this is good to hear, and this demonstrates the power of using more data.
The only issue here, as I mentioned before, is that in joint XN refinement we refine one single model against two data sets, while ideally it should be two models in order to account for genuine unique information that is contained in X-ray and neutron data. This will be addressed at some point.
Out of curiosity, could you elaborate on why it would be preferred to refine two separate models? It seems to me, admittedly not having any experience with neutron diffraction, that the ultimate goal is to model the system of the crystal, and given 2 data sets, we would want to create the single model describing the structure in the crystal that best agrees with all the available data. Are the neutron and X-ray data so different that using them for joint refinement is disadvantageous?
I think I wrote that already in previous email.. Structures corresponding to two data sets are of course overall similar may be different in local details: - Different X-H(D) distances (see figure 4 here: http://journals.iucr.org/d/issues/2010/11/00/dz5209/index.html) - Different data collection temperatures: -- Different B factors; -- Different visible water structure; -- Different local disorder (that may be captured as distinct conformations at low temperature, but may be blurred at higher temperature); - Different completeness and resolution of one data set may allow features in one model corresponding to this data set that may not be supported by the other data set. For example, you may see a well resolved water with one or two D in neutron map but you may not even see any water at all in X-ray map, or the other way around. - etc etc etc.
So it is advantageous to use both X-ray and neutron data sets together but these nuances need to be carefully taken care of which isn't the case in current implementations of joint XN refinement protocols. This makes a good room for future improvements.
Pavel _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]
-- Professor John R Helliwell DSc
Hi Pavel, as I understand, phenix is alrady on route for the 'future improvements' you point out below. That's great news, and an excellent idea to internally have two models for two data sets - that will give phenix much better fine-tuning options for conventional joint refinement or using the X-ray model as reference model. It sounds like a very promising approach. Best, Tim On Saturday, October 15, 2016 11:40:49 AM Pavel Afonine wrote: > >> this is good to hear, and this demonstrates the power of using more data. > >> The only issue here, as I mentioned before, is that in joint XN > >> refinement we refine one single model against two data sets, while > >> ideally it should be two models in order to account for genuine unique > >> information that is contained in X-ray and neutron data. This will be > >> addressed at some point.> > > Out of curiosity, could you elaborate on why it would be preferred to > > refine two separate models? It seems to me, admittedly not having any > > experience with neutron diffraction, that the ultimate goal is to model > > the system of the crystal, and given 2 data sets, we would want to create > > the single model describing the structure in the crystal that best agrees > > with all the available data. Are the neutron and X-ray data so different > > that using them for joint refinement is disadvantageous? > I think I wrote that already in previous email.. Structures > corresponding to two data sets are of course overall similar may be > different in local details: > - Different X-H(D) distances (see figure 4 here: > http://journals.iucr.org/d/issues/2010/11/00/dz5209/index.html) > - Different data collection temperatures: > -- Different B factors; > -- Different visible water structure; > -- Different local disorder (that may be captured as distinct > conformations at low temperature, but may be blurred at higher temperature); > - Different completeness and resolution of one data set may allow > features in one model corresponding to this data set that may not be > supported by the other data set. For example, you may see a well > resolved water with one or two D in neutron map but you may not even see > any water at all in X-ray map, or the other way around. > - etc etc etc. > > So it is advantageous to use both X-ray and neutron data sets together > but these nuances need to be carefully taken care of which isn't the > case in current implementations of joint XN refinement protocols. This > makes a good room for future improvements. > > Pavel > _______________________________________________ > phenixbb mailing list > [email protected] > http://phenix-online.org/mailman/listinfo/phenixbb > Unsubscribe: [email protected] -- -- Paul Scherrer Institut Dr. Tim Gruene - persoenlich - Principal Investigator Biology and Chemistry OFLC/102 CH-5232 Villigen PSI Phone: +41 (0)56 310 5297 GPG Key ID = A46BEE1A
participants (3)
-
John R Helliwell -
Pavel Afonine -
Tim Gruene