Re: [ccp4bb] cell discrepancies and stuck refinement using different XDS-versions
>>But there is no rule without exception, Well, occasionally there is. JPK To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB=1
Re: [ccp4bb] cell discrepancies and stuck refinement using different XDS-versions
Dear all,
I post this again to CCP4BB because some people informed me that they
received an empty email from me. This time I use my email client;
previously I used the Jiscmail web interface - does anybody know what
went wrong? Hopefully it works this time. Sorry if you get this twice.
Kay
Dear Sabine,
indeed, the differences between the last XDS versions are in the area of
indexing, with BUILT=20180409 being the most robust (and recommended)
one. This gives the desired results in all my tests, and GlobalPhasing
has successfully tested it with ~60 difficult cases. But there is no
rule without exception, which means that it could nevertheless give the
wrong answer in your case.
According to CCP4 othercell, the two cells that you cite are related by
the [h-2l,-h,k] reindexing operator. However the cell volumes differ by
slightly more than a factor of 2 , namely 232274 / 110672 = 2.1 which I
find astonishing because typically such different indexing results
result in a pseudo-centering situation, where the indexing results
differ in that one tries to explain the weak and strong reflections,
whereas the other only explains the strong reflections. What exactly
happens in your case I don't know; I'd have to see the data and the
detailed output of XDS. It could also be a case of two lattices (from
two crystals) superimposed.
You can probably get the small cell with BUILT=20180409 if you either
a) specify it in XDS.INP (together with SPACE_GROUP_NUMBER=1)
b) or remove the weak reflections from SPOT.XDS
c) or by adjusting some parameters in XDS.INP
To really understand what is going on in your case, you could
a) run the checkcentering program (from
ftp://turn5.biologie.uni-konstanz.de/pub/linux_bin ) which will analyze
your data w.r.t. pseudo-centering (similar to what pointless does)
b) look, in a representation of reciprocal space, at the spots that are
used for indexing and that are indexed, or not. For this, you need the
spot2pdb program (same download directory). Run it
grep -s allow-duplicate-sequence-numbers ~/.coot || echo
"(allow-duplicate-sequence-numbers)" >>~/.coot
spot2pdb
coot SPOT-*.pdb
as suggested in
https://strucbio.biologie.uni-konstanz.de/xdswiki/index.php/XDSGUI#tools
BTW it is normal that R factors with tNCS are higher ("stuck") than
without; this is due to the wider distribution of reflection intensities
in the case of tNCS (many weak ones, and many strong ones; less with
intermediate intensity). Most people would say that the space group and
cell where the refinement proceeds best must be the correct one; however
if that does not explain all the reflections of the lattice then this is
not quite satisfactory.
best wishes,
Kay
On Mon, 6 Aug 2018 16:50:56 +0200, Sabine Schneider
wrote:
>Dear all,
>
>We are encountering a differences in indexing using different XDS
>versions, which results in either 2 or 4 mols/asu (SG P1; structure
>determination via MR (30% seq ID model)) and either successful
>refinement or stuck R/Rfree
>
>- the data were collected at ESRF ID30A3 (Aiger detector) and extend to
>about 2.3A resolution (CC1/2 ~50%)
>
>The XDS version build 20180126 (and or autoprocessing at the ESRF via
>autoproc, dials, xdsapp or grenade -> here also xds version from
>20180126 used) gives us:
>
>P1 38.6550.7661.11 110.057 99.945 90.197
>XDS complains and I need to use "DEFPIX INTEGRATE CORRECT"
>-> 2mols/asu, structure refines to R/Rfree of 22/25
>
>In contrast the actual XDS-Version BUILT=20180409 results in:
>P1 39.2 51.3 116.8 86.3 82.3 89.8
>but XDS runs smoothly.
>-> 4mols/asu, tNCS, R/Rfree stuck at 28/32
>If I feed XDS with the smaller cell above, it fails.
>
>(The smaller cell is also found by Xia2/dials via the CCP4i2 interface.)
>
>Thus I am wondering, what are the differences between the two
>XDS-versions? (I remember vaguely that there was a tread about different
>XDS-versions, but couldn't find it..)
>
>Cheers Sabine
>
>
>--
>--
>Dr. Sabine Schneider
>Research Group Leader
>Technical University of Munich
>Department of Chemistry
>Chair of Biochemistry
>Lichtenbergstr. 4
>85748 Garching
>Germany
>Tel.: +49 (0) 89 289 13759
>Fax: +49 (0) 89 289 13363
>http://www.biochemie.ch.tum.de/index.php?id=919
>
>--
>--
>Dr. Sabine Schneider
>Research Group Leader
>Technical University of Munich
>Department of Chemistry
>Chair of Biochemistry
>Lichtenbergstr. 4
>85748 Garching
>Germany
>Tel.: +49 (0) 89 289 13759
>Fax: +49 (0) 89 289 13363
>http://www.biochemie.ch.tum.de/index.php?id=919
>
>
>
>To unsubscribe from the CCP4BB list, click the following link:
>https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB=1
--
Kay Diederichshttp://strucbio.biologie.uni-konstanz.de
email: kay.diederi...@uni-konstanz.deTel +49 7531 88 4049 Fax 3183
Fachbereich
Re: [ccp4bb] cell discrepancies and stuck refinement using different XDS-versions
Dear Sabine,
indeed, the differences between the last XDS versions are in the area of
indexing, with BUILT=20180409 being the most robust (and recommended) one.
This gives the desired results in all my tests, and GlobalPhasing has
successfully tested it with ~60 difficult cases. But there is no rule without
exception, which means that it could nevertheless give the wrong answer in your
case.
According to CCP4 othercell, the two cells that you cite are related by the
[h-2l,-h,k] reindexing operator. However the cell volumes differ by slightly
more than a factor of 2 , namely 232274 / 110672 = 2.1 which I find astonishing
because typically such different indexing results result in a pseudo-centering
situation, where the indexing results differ in that one tries to explain the
weak and strong reflections, whereas the other only explains the strong
reflections. What exactly happens in your case I don't know; I'd have to see
the data and the detailed output of XDS. It could also be a case of two
lattices (from two crystals) superimposed.
You can probably get the small cell with BUILT=20180409 if you either
a) specify it in XDS.INP (together with SPACE_GROUP_NUMBER=1)
b) or remove the weak reflections from SPOT.XDS
c) or by adjusting some parameters in XDS.INP
To really understand what is going on in your case, you could
a) run the checkcentering program (from
ftp://turn5.biologie.uni-konstanz.de/pub/linux_bin ) which will analyze your
data w.r.t. pseudo-centering (similar to what pointless does)
b) look, in a representation of reciprocal space, at the spots that are used
for indexing and that are indexed, or not. For this, you need the spot2pdb
program (same download directory). Run it
grep -s allow-duplicate-sequence-numbers ~/.coot || echo
"(allow-duplicate-sequence-numbers)" >>~/.coot
spot2pdb
coot SPOT-*.pdb
as suggested in
https://strucbio.biologie.uni-konstanz.de/xdswiki/index.php/XDSGUI#tools
BTW it is normal that R factors with tNCS are higher ("stuck") than without;
this is due to the wider distribution of reflection intensities in the case of
tNCS (many weak ones, and many strong ones; less with intermediate intensity).
Most people would say that the space group and cell where the refinement
proceeds best must be the correct one; however if that does not explain all the
reflections of the lattice then this is not quite satisfactory.
best wishes,
Kay
On Mon, 6 Aug 2018 16:50:56 +0200, Sabine Schneider
wrote:
>Dear all,
>
>We are encountering a differences in indexing using different XDS
>versions, which results in either 2 or 4 mols/asu (SG P1; structure
>determination via MR (30% seq ID model)) and either successful
>refinement or stuck R/Rfree
>
>- the data were collected at ESRF ID30A3 (Aiger detector) and extend to
>about 2.3A resolution (CC1/2 ~50%)
>
>The XDS version build 20180126 (and or autoprocessing at the ESRF via
>autoproc, dials, xdsapp or grenade -> here also xds version from
>20180126 used) gives us:
>
>P1 38.65 50.76 61.11 110.057 99.945 90.197
>XDS complains and I need to use "DEFPIX INTEGRATE CORRECT"
>-> 2mols/asu, structure refines to R/Rfree of 22/25
>
>In contrast the actual XDS-Version BUILT=20180409 results in:
>P1 39.2 51.3 116.8 86.3 82.3 89.8
>but XDS runs smoothly.
>-> 4mols/asu, tNCS, R/Rfree stuck at 28/32
>If I feed XDS with the smaller cell above, it fails.
>
>(The smaller cell is also found by Xia2/dials via the CCP4i2 interface.)
>
>Thus I am wondering, what are the differences between the two
>XDS-versions? (I remember vaguely that there was a tread about different
>XDS-versions, but couldn't find it..)
>
>Cheers Sabine
>
>
>--
>--
>Dr. Sabine Schneider
>Research Group Leader
>Technical University of Munich
>Department of Chemistry
>Chair of Biochemistry
>Lichtenbergstr. 4
>85748 Garching
>Germany
>Tel.: +49 (0) 89 289 13759
>Fax: +49 (0) 89 289 13363
>http://www.biochemie.ch.tum.de/index.php?id=919
>
>--
>--
>Dr. Sabine Schneider
>Research Group Leader
>Technical University of Munich
>Department of Chemistry
>Chair of Biochemistry
>Lichtenbergstr. 4
>85748 Garching
>Germany
>Tel.: +49 (0) 89 289 13759
>Fax: +49 (0) 89 289 13363
>http://www.biochemie.ch.tum.de/index.php?id=919
>
>
>
>To unsubscribe from the CCP4BB list, click the following link:
>https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB=1
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB=1
Re: [ccp4bb] cell discrepancies and stuck refinement using different XDS-versions
Dear Sabine, I would first take a look at the respective log-files for differences in parameter settings, in particular SPOT_RANGE / DATA_RANGE, INCLUDE_RESOLUTION_RANGE. Start with IDXREF.LP, then proceed to COLSPOT.LP. The relevant parameters are always listed in the header of the log-files. Programms like xxdiff or vimdiff are very useful to highlight the differences. Best, Tim On Monday, August 6, 2018 4:50:56 PM CEST Sabine Schneider wrote: > Dear all, > > We are encountering a differences in indexing using different XDS > versions, which results in either 2 or 4 mols/asu (SG P1; structure > determination via MR (30% seq ID model)) and either successful > refinement or stuck R/Rfree > > - the data were collected at ESRF ID30A3 (Aiger detector) and extend to > about 2.3A resolution (CC1/2 ~50%) > > The XDS version build 20180126 (and or autoprocessing at the ESRF via > autoproc, dials, xdsapp or grenade -> here also xds version from > 20180126 used) gives us: > > P1 38.6550.7661.11 110.057 99.945 90.197 > XDS complains and I need to use "DEFPIX INTEGRATE CORRECT" > -> 2mols/asu, structure refines to R/Rfree of 22/25 > > In contrast the actual XDS-Version BUILT=20180409 results in: > P1 39.2 51.3 116.8 86.3 82.3 89.8 > but XDS runs smoothly. > -> 4mols/asu, tNCS, R/Rfree stuck at 28/32 > If I feed XDS with the smaller cell above, it fails. > > (The smaller cell is also found by Xia2/dials via the CCP4i2 interface.) > > Thus I am wondering, what are the differences between the two > XDS-versions? (I remember vaguely that there was a tread about different > XDS-versions, but couldn't find it..) > > Cheers Sabine -- -- Paul Scherrer Institut Tim Gruene - persoenlich - OSUA/204 CH-5232 Villigen PSI phone: +41 (0)56 310 5297 GPG Key ID = A46BEE1A To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB=1 signature.asc Description: This is a digitally signed message part.
[ccp4bb] cell discrepancies and stuck refinement using different XDS-versions
Dear all, We are encountering a differences in indexing using different XDS versions, which results in either 2 or 4 mols/asu (SG P1; structure determination via MR (30% seq ID model)) and either successful refinement or stuck R/Rfree - the data were collected at ESRF ID30A3 (Aiger detector) and extend to about 2.3A resolution (CC1/2 ~50%) The XDS version build 20180126 (and or autoprocessing at the ESRF via autoproc, dials, xdsapp or grenade -> here also xds version from 20180126 used) gives us: P1 38.65 50.76 61.11 110.057 99.945 90.197 XDS complains and I need to use "DEFPIX INTEGRATE CORRECT" -> 2mols/asu, structure refines to R/Rfree of 22/25 In contrast the actual XDS-Version BUILT=20180409 results in: P1 39.2 51.3 116.8 86.3 82.3 89.8 but XDS runs smoothly. -> 4mols/asu, tNCS, R/Rfree stuck at 28/32 If I feed XDS with the smaller cell above, it fails. (The smaller cell is also found by Xia2/dials via the CCP4i2 interface.) Thus I am wondering, what are the differences between the two XDS-versions? (I remember vaguely that there was a tread about different XDS-versions, but couldn't find it..) Cheers Sabine -- -- Dr. Sabine Schneider Research Group Leader Technical University of Munich Department of Chemistry Chair of Biochemistry Lichtenbergstr. 4 85748 Garching Germany Tel.: +49 (0) 89 289 13759 Fax: +49 (0) 89 289 13363 http://www.biochemie.ch.tum.de/index.php?id=919 -- -- Dr. Sabine Schneider Research Group Leader Technical University of Munich Department of Chemistry Chair of Biochemistry Lichtenbergstr. 4 85748 Garching Germany Tel.: +49 (0) 89 289 13759 Fax: +49 (0) 89 289 13363 http://www.biochemie.ch.tum.de/index.php?id=919 To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB=1
