Re: [ccp4bb] arp/warp in p22121: what to do in Pointless
Hi Phil, Just in case anyone was used to the old default behaviour (and like to use the current version of Arp/wArp ;oD) could you ensure that there is an option which will allow the automatic reindexing to the old default setting? Thanks, Graeme From: CCP4 bulletin board on behalf of Phil Evans Sent: Mon 24/09/2007 6:18 PM To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] arp/warp in p22121: what to do in Pointless OK I'll try to change the default behaviour, after I've fixed the current round of bugs! Sometimes reindexing is required from the Mosflm indexing even without a reference set, if the Laue group assignment is wrong, or where there is pseudo lattice symmetry (eg we have a case which is really R3 but pseudo I-centred cubic, so Mosflm may choose the wrong 3-fold in its indexing) My POV was that on the first dataset, you don't care about the indexing, just about the space group, but this is undermined by the frequent uncertainty about systematic absences. I think the best thing to do is to choose the reindexing for the Laue group (or point group), and not to switch it for different possible space groups within the Laue group. I think this should be able to pick up the I121 setting of C2 (though I might have to make that a special case: similarly George's P21/n v. P21/c) Phil On 24 Sep 2007, at 17:59, Ian Tickle wrote: > > Phil, > > It also affects centred monoclinic: to avoid some cases of beta > 120 > you have to use the I121 setting instead of C121 (and it is a > conventional setting, see IT vol. A pp.126-7). For example the > conventional (but non-standard) I121 setting: a=50 b=60 c=51 beta=90.2 > would I think be re-indexed by pointless to the unconventional (but > standard) C121 setting: a=71.3 b=60 c=50 beta=134.3. > > Trigonal, tetragonal & hexagonal settings with a or b unique are > definitely not only non-standard but also non-conventional: they > violate > Table 9.3.4.1 in IT Vol. A and they don't appear in the diagrams, so > let's not go there! The point about the IUCr convention is that the > cell is determined by the rules in Table 9.3.4.1 so you would never > get > into that situation. For the oP lattice the cell is determined > once and > for all by the rules in exactly the same way: the systematic absences > are not taken into account because they are not always reliable, > e.g. it > may be only after you've solved the TF that you can assign the SG > correctly, or at least beyond reasonable doubt, and re-indexing at > that > point makes no sense. > > To summarise my POV: I don't really care which convention you use, > but I > think gratuitous re-indexing (i.e. without the user specifically > asking > for it) is just very confusing, and this is what can happen if you > stick > to the standard settings. I'm all for educating users, and re- > indexing > 1 or 2 datasets can be very educational, but if you have lots of > datasets to deal with it's no longer amusing! Sometimes re-indexing > *is* needed, e.g. because you already have isomorphous protein-ligand > structures and you want to be able to compare them easily, but this > should *only* be done to conform to a reference dataset. I don't see > any point in re-indexing just to conform to the 'standard setting' > 'non-convention' which is derived from a mis-reading of the 1952 ed of > IT (my guess is they didn't list all the conventional settings because > the publisher wanted to save money! - but at least they saw sense for > the 1983 ed.). > > In any case this will only affect users of Arp/Warp, all other CCP4 > programs (and most non-CCP4 programs, e.g. Shel-X, CNS etc) can handle > non-standard settings, and if you use Mosflm and always specify SG > P222 > for all oP cases (which I always do anyway), you would have to re- > index > anyway. > > -- Ian > >> -Original Message- >> From: [EMAIL PROTECTED] >> [mailto:[EMAIL PROTECTED] On Behalf Of Phil Evans >> Sent: 24 September 2007 16:58 >> To: CCP4BB@JISCMAIL.AC.UK >> Subject: Re: [ccp4bb] arp/warp in p22121: what to do in Pointless >> >> I think this is only a problem in the primitive orthorhombic system >> (at least I assume people don't want hexagonal axes along a, A & B >> centred lattices etc, although there is no reason in >> principle why not). >> >> Following some earlier discussions with Ian, Pointless now honours >> (and preserves) a reference file (HKLREF) in eg P 2 21 21, and also >> explicit reindex operations, but an initial indexing will still >> enforce the "standard" setting eg P 21 21 2, because I accept the >>
Re: [ccp4bb] arp/warp in p22121: what to do in Pointless
OK I'll try to change the default behaviour, after I've fixed the current round of bugs! Sometimes reindexing is required from the Mosflm indexing even without a reference set, if the Laue group assignment is wrong, or where there is pseudo lattice symmetry (eg we have a case which is really R3 but pseudo I-centred cubic, so Mosflm may choose the wrong 3-fold in its indexing) My POV was that on the first dataset, you don't care about the indexing, just about the space group, but this is undermined by the frequent uncertainty about systematic absences. I think the best thing to do is to choose the reindexing for the Laue group (or point group), and not to switch it for different possible space groups within the Laue group. I think this should be able to pick up the I121 setting of C2 (though I might have to make that a special case: similarly George's P21/n v. P21/c) Phil On 24 Sep 2007, at 17:59, Ian Tickle wrote: Phil, It also affects centred monoclinic: to avoid some cases of beta > 120 you have to use the I121 setting instead of C121 (and it is a conventional setting, see IT vol. A pp.126-7). For example the conventional (but non-standard) I121 setting: a=50 b=60 c=51 beta=90.2 would I think be re-indexed by pointless to the unconventional (but standard) C121 setting: a=71.3 b=60 c=50 beta=134.3. Trigonal, tetragonal & hexagonal settings with a or b unique are definitely not only non-standard but also non-conventional: they violate Table 9.3.4.1 in IT Vol. A and they don't appear in the diagrams, so let's not go there! The point about the IUCr convention is that the cell is determined by the rules in Table 9.3.4.1 so you would never get into that situation. For the oP lattice the cell is determined once and for all by the rules in exactly the same way: the systematic absences are not taken into account because they are not always reliable, e.g. it may be only after you've solved the TF that you can assign the SG correctly, or at least beyond reasonable doubt, and re-indexing at that point makes no sense. To summarise my POV: I don't really care which convention you use, but I think gratuitous re-indexing (i.e. without the user specifically asking for it) is just very confusing, and this is what can happen if you stick to the standard settings. I'm all for educating users, and re- indexing 1 or 2 datasets can be very educational, but if you have lots of datasets to deal with it's no longer amusing! Sometimes re-indexing *is* needed, e.g. because you already have isomorphous protein-ligand structures and you want to be able to compare them easily, but this should *only* be done to conform to a reference dataset. I don't see any point in re-indexing just to conform to the 'standard setting' 'non-convention' which is derived from a mis-reading of the 1952 ed of IT (my guess is they didn't list all the conventional settings because the publisher wanted to save money! - but at least they saw sense for the 1983 ed.). In any case this will only affect users of Arp/Warp, all other CCP4 programs (and most non-CCP4 programs, e.g. Shel-X, CNS etc) can handle non-standard settings, and if you use Mosflm and always specify SG P222 for all oP cases (which I always do anyway), you would have to re- index anyway. -- Ian -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Phil Evans Sent: 24 September 2007 16:58 To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] arp/warp in p22121: what to do in Pointless I think this is only a problem in the primitive orthorhombic system (at least I assume people don't want hexagonal axes along a, A & B centred lattices etc, although there is no reason in principle why not). Following some earlier discussions with Ian, Pointless now honours (and preserves) a reference file (HKLREF) in eg P 2 21 21, and also explicit reindex operations, but an initial indexing will still enforce the "standard" setting eg P 21 21 2, because I accept the "reference" setting from the cctbx library ie suppose you have a crystal which when indexed with a <= b <= c and Pointless decides unambiguously for the sake of argument) that the axis along a is a 2-fold and the other two are 2(1) screws, ie space group P 2 21 21. At present this will be reindexed to the "standard" setting P 21 21 2, but is that what you want, or should it be left as a Hi Sue It's certainly true that the convention in the 1935 and 1952 editions of IT Volume 1 *appeared* to be the 'standard setting' convention that you describe because only the 'standard' settings were listed, and this was the way that many crystallographers interpreted it (actually only macromolecular crystallographers, the small molecule people stick to the IUCr convention), so this is prob
Re: [ccp4bb] arp/warp in p22121: what to do in Pointless
Phil, It also affects centred monoclinic: to avoid some cases of beta > 120 you have to use the I121 setting instead of C121 (and it is a conventional setting, see IT vol. A pp.126-7). For example the conventional (but non-standard) I121 setting: a=50 b=60 c=51 beta=90.2 would I think be re-indexed by pointless to the unconventional (but standard) C121 setting: a=71.3 b=60 c=50 beta=134.3. Trigonal, tetragonal & hexagonal settings with a or b unique are definitely not only non-standard but also non-conventional: they violate Table 9.3.4.1 in IT Vol. A and they don't appear in the diagrams, so let's not go there! The point about the IUCr convention is that the cell is determined by the rules in Table 9.3.4.1 so you would never get into that situation. For the oP lattice the cell is determined once and for all by the rules in exactly the same way: the systematic absences are not taken into account because they are not always reliable, e.g. it may be only after you've solved the TF that you can assign the SG correctly, or at least beyond reasonable doubt, and re-indexing at that point makes no sense. To summarise my POV: I don't really care which convention you use, but I think gratuitous re-indexing (i.e. without the user specifically asking for it) is just very confusing, and this is what can happen if you stick to the standard settings. I'm all for educating users, and re-indexing 1 or 2 datasets can be very educational, but if you have lots of datasets to deal with it's no longer amusing! Sometimes re-indexing *is* needed, e.g. because you already have isomorphous protein-ligand structures and you want to be able to compare them easily, but this should *only* be done to conform to a reference dataset. I don't see any point in re-indexing just to conform to the 'standard setting' 'non-convention' which is derived from a mis-reading of the 1952 ed of IT (my guess is they didn't list all the conventional settings because the publisher wanted to save money! - but at least they saw sense for the 1983 ed.). In any case this will only affect users of Arp/Warp, all other CCP4 programs (and most non-CCP4 programs, e.g. Shel-X, CNS etc) can handle non-standard settings, and if you use Mosflm and always specify SG P222 for all oP cases (which I always do anyway), you would have to re-index anyway. -- Ian > -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Phil Evans > Sent: 24 September 2007 16:58 > To: CCP4BB@JISCMAIL.AC.UK > Subject: Re: [ccp4bb] arp/warp in p22121: what to do in Pointless > > I think this is only a problem in the primitive orthorhombic system > (at least I assume people don't want hexagonal axes along a, A & B > centred lattices etc, although there is no reason in > principle why not). > > Following some earlier discussions with Ian, Pointless now honours > (and preserves) a reference file (HKLREF) in eg P 2 21 21, and also > explicit reindex operations, but an initial indexing will still > enforce the "standard" setting eg P 21 21 2, because I accept the > "reference" setting from the cctbx library > > ie suppose you have a crystal which when indexed with a <= b > <= c and > Pointless decides unambiguously for the sake of argument) that the > axis along a is a 2-fold and the other two are 2(1) screws, ie space > group P 2 21 21. > > At present this will be reindexed to the "standard" setting P 21 21 > 2, but is that what you want, or should it be left as a criterion takes precedence? > > Phil > > > On 19 Sep 2007, at 17:54, Ian Tickle wrote: > > > Hi Sue > > > > It's certainly true that the convention in the 1935 and 1952 > > editions of > > IT Volume 1 *appeared* to be the 'standard setting' > convention that > > you > > describe because only the 'standard' settings were listed, > and this > > was > > the way that many crystallographers interpreted it (actually only > > macromolecular crystallographers, the small molecule people stick > > to the > > IUCr convention), so this is probably where you're coming from. > > However > > the 1983 edition of Volume A clarified the situation and > made it clear > > that this was never the intention, so all the conventional > settings > > are > > now shown on the SG diagram pages. P22121 & P21221 certainly are > > defined in IT Vol. A - look on the diagram page for SG no. > 18 & you'll > > see them. > > > > The 'standard symbol' for a space group is merely the heading on the > > page used only for indexing purposes, so space groups P22121, >
Re: [ccp4bb] arp/warp in p22121: what to do in Pointless
Phil, For the record, the program XPREP - widely used by small molecule crystallographers but also useful for macromoleculess - always makes the conventional cell (in this example P 21 21 2) the default (i.e. what you get if you always hit ), and writes out new .hkl and .ins files in this setting unless the user forces it to do otherwise (the .ins file is used for solving the structure with direct methods and contains the cell corresponding to the new .hkl file). There are three cases where it is intentionally made less awkward for the user to choose an alternative setting if she/he so wishes: 1. Primitive rhombohedral rather than the hexagonal cell with three times the volume. Unlike the situation for macromolecules, where some older programs do not work with the primitive rhombohedral cell, the primitive cell is quite often chosen for small molecules. 2. The space group I2 rather than C2 if it results in a beta angle much closer to 90 degrees (or when it is desired to compare the structure with a structure - possibly another crystalline phase of the same compound - in say I222). The same applies to other C-centred monoclinic cells, e.g. Cc <-> Ia etc. (but Fd is discouraged). 3. The setting P21/n rather than P21/c if it results in a beta angles much closer to 90 etc. These three 'unconventional' settings are also generally accepted by the checking software, databases and journals. To cover all reasonable non-standard settings, XPREP actually understands 413 different space groups! Best wishes, George Prof. George M. Sheldrick FRS Dept. Structural Chemistry, University of Goettingen, Tammannstr. 4, D37077 Goettingen, Germany Tel. +49-551-39-3021 or -3068 Fax. +49-551-39-2582 On Mon, 24 Sep 2007, Phil Evans wrote: > I think this is only a problem in the primitive orthorhombic system (at least > I assume people don't want hexagonal axes along a, A & B centred lattices etc, > although there is no reason in principle why not). > > Following some earlier discussions with Ian, Pointless now honours (and > preserves) a reference file (HKLREF) in eg P 2 21 21, and also explicit > reindex operations, but an initial indexing will still enforce the "standard" > setting eg P 21 21 2, because I accept the "reference" setting from the cctbx > library > > ie suppose you have a crystal which when indexed with a <= b <= c and > Pointless decides unambiguously for the sake of argument) that the axis along > a is a 2-fold and the other two are 2(1) screws, ie space group P 2 21 21. > > At present this will be reindexed to the "standard" setting P 21 21 2, but is > that what you want, or should it be left as a precedence? > > Phil > > > On 19 Sep 2007, at 17:54, Ian Tickle wrote: > > > Hi Sue > > > > It's certainly true that the convention in the 1935 and 1952 editions of > > IT Volume 1 *appeared* to be the 'standard setting' convention that you > > describe because only the 'standard' settings were listed, and this was > > the way that many crystallographers interpreted it (actually only > > macromolecular crystallographers, the small molecule people stick to the > > IUCr convention), so this is probably where you're coming from. However > > the 1983 edition of Volume A clarified the situation and made it clear > > that this was never the intention, so all the conventional settings are > > now shown on the SG diagram pages. P22121 & P21221 certainly are > > defined in IT Vol. A - look on the diagram page for SG no. 18 & you'll > > see them. > > > > The 'standard symbol' for a space group is merely the heading on the > > page used only for indexing purposes, so space groups P22121, P21221 and > > P21212 all have the same standard symbol P21212; hence the standard > > symbol is not unique and can't be used to unambiguously define the space > > group. The 'standard setting' is merely the space group setting that > > has the same name as the standard symbol. Even if that weren't true do > > we really want to be still sticking to a convention that was abandoned > > 25 years ago and doesn't a later convention override an earlier one > > anyway? > > > > Actually the convention in use is not the issue anyway, I don't care > > which convention is used as long as all programs use the same > > convention! - then I'll never need to permute axes (just as > > fundamentally I don't care which co-ordinate format is used as long as > > all programs use the same one, then I'll never need to reformat). So > > Mosflm uses the IUCr convention (i.e. a<=b<=c for primitive > > orthorhombic), and therefore any program which doesn't support that > > convention for any space group forces you to permute the axes completely > > unnecessarily. > > > > -- Ian > > > > > -Original Message- > > > From: Sue Roberts [mailto:[EMAIL PROTECTED] > > > Sent: 19 September 2007 16:38 > > > To: Ian Tickle > > > Subject: Re: [ccp4bb] arp/warp in p22121 > > > > > > Hi Ian > > > > > > But there's an older co
Re: [ccp4bb] arp/warp in p22121: what to do in Pointless
I think this is only a problem in the primitive orthorhombic system (at least I assume people don't want hexagonal axes along a, A & B centred lattices etc, although there is no reason in principle why not). Following some earlier discussions with Ian, Pointless now honours (and preserves) a reference file (HKLREF) in eg P 2 21 21, and also explicit reindex operations, but an initial indexing will still enforce the "standard" setting eg P 21 21 2, because I accept the "reference" setting from the cctbx library ie suppose you have a crystal which when indexed with a <= b <= c and Pointless decides unambiguously for the sake of argument) that the axis along a is a 2-fold and the other two are 2(1) screws, ie space group P 2 21 21. At present this will be reindexed to the "standard" setting P 21 21 2, but is that what you want, or should it be left as acriterion takes precedence? Phil On 19 Sep 2007, at 17:54, Ian Tickle wrote: Hi Sue It's certainly true that the convention in the 1935 and 1952 editions of IT Volume 1 *appeared* to be the 'standard setting' convention that you describe because only the 'standard' settings were listed, and this was the way that many crystallographers interpreted it (actually only macromolecular crystallographers, the small molecule people stick to the IUCr convention), so this is probably where you're coming from. However the 1983 edition of Volume A clarified the situation and made it clear that this was never the intention, so all the conventional settings are now shown on the SG diagram pages. P22121 & P21221 certainly are defined in IT Vol. A - look on the diagram page for SG no. 18 & you'll see them. The 'standard symbol' for a space group is merely the heading on the page used only for indexing purposes, so space groups P22121, P21221 and P21212 all have the same standard symbol P21212; hence the standard symbol is not unique and can't be used to unambiguously define the space group. The 'standard setting' is merely the space group setting that has the same name as the standard symbol. Even if that weren't true do we really want to be still sticking to a convention that was abandoned 25 years ago and doesn't a later convention override an earlier one anyway? Actually the convention in use is not the issue anyway, I don't care which convention is used as long as all programs use the same convention! - then I'll never need to permute axes (just as fundamentally I don't care which co-ordinate format is used as long as all programs use the same one, then I'll never need to reformat). So Mosflm uses the IUCr convention (i.e. a<=b<=c for primitive orthorhombic), and therefore any program which doesn't support that convention for any space group forces you to permute the axes completely unnecessarily. -- Ian -Original Message- From: Sue Roberts [mailto:[EMAIL PROTECTED] Sent: 19 September 2007 16:38 To: Ian Tickle Subject: Re: [ccp4bb] arp/warp in p22121 Hi Ian But there's an older convention, which is to use the space groups settings defined in the International Tables - and P22121 is not a standard setting. Sue On Sep 19, 2007, at 8:18 AM, Ian Tickle wrote: I'm confused now, sticking to the IUCr convention should not require any axis permutation. My beef is specifically against unnecessary axis permutations! Surely it's when the program doesn't support the convention that you are forced to permute the axes? Besides I did solve a structure in P22121 with Phaser so I'm even more confused! -- Ian -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Airlie McCoy Sent: 19 September 2007 15:09 To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] arp/warp in p22121 The problem is specifically that ARP/wARP *doesn't* support the IUCr convention as given in IT (Vol. A, >= 1983 edition, Table 9.3.4.1, p.758 in 5th ed.) regarding choice of cell in primitive orthorhombic space groups, and I suspect in centred monoclinic ones also. AFAIK ARP/wARP and pointless are the only two CCP4 programs that currently don't fully support the IUCr convention Phaser doesn't "support" the IUCr convention, and if it was used for the original MR in this case (I don't know whether it was or not), then it would have caused the "problem". We have had user requests to change the output to the IUCr convention, but other people get confused if the axes are permuted. So the choice will be made an output option - Frank von Delft suggested the keyword "IUCR [ON/OFF]"! Vote for your choice of default now... Airlie McCoy Disclaimer This communication is confidential and may contain privileged information intended solely for the named addressee(s). It may not be used or disclosed except for the purpose for which it has been sent. If you are not the intended recipient you must not review, use, disclose, copy, distribute or take any action in reliance upon it. If you have r