I noticed that my response didn't go through due to the attachment being to
big. The PDB file is here:
https://dl.dropboxusercontent.com/u/41995424/4BVNAa-2Z73Aa.pdb
-Spencer
On Thu, Nov 6, 2014 at 2:44 PM, Spencer Bliven <sbli...@ucsd.edu> wrote:
> I hadn't heard about any plans to convert biological assemblies to mmcif
> format, although that would certainly be a logical move. As far as I know,
> only split entries will be discontinued in December. I'm not sure if even
> mmCIF has a clean solution to the problem that we would sometimes like to
> display and access multiple copies of the asymmetric unit in different
> positions. The issue is broader than just the assembly files, and it is
> also a problem when dealing with PPIs or when doing structural comparisons.
>
> I've attached a PDB file with two non-identical states. It was saved from
> the RCSB protein comparison tool (
> http://rcsb.org/pdb/workbench/getaligned.do?action=pw_fatcat&mol=PDP:4BVNAa&mol=PDP:2Z73Aa).
> Of course, that doesn't mean that this use of MODEL was blessed by the
> wwPDB format committee, but it is a useful hack to save multiple objects in
> a single PDB file (and works great in Jmol).
>
> -Spencer
>
>
> On Thu, Nov 6, 2014 at 12:51 PM, David Hall <li...@cowsandmilk.net> wrote:
>
>> I would tell you to use mmcif (especially since the structures I believe
>> you are describing will no longer be available in pdb format in a month),
>> but it might not solve part of your problem. I'm still not sure what PDB id
>> has a chain in state 1 that does not have exactly the same internal
>> coordinates as the chain in state 2 though, can you provide an example.
>>
>> On Thu, Nov 6, 2014 at 6:19 AM, Spencer Bliven <sbli...@ucsd.edu> wrote:
>>
>>> One possible reason for this behavior would be code that assumes atom
>>> IDs are unique. In fact, both the ID and the state are required to uniquely
>>> identify an atom. This would be consistent with some other problems I've
>>> observed when working with multi-state objects, such as the fact that
>>> cartoon representation does not work on states which don't match the atoms
>>> of state 1. Does anyone familiar with the code know if that's the case? If
>>> so it's probably not a trivial fix.
>>>
>>> -Spencer
>>>
>>> On Thu, Nov 6, 2014 at 10:18 AM, Spencer Bliven <sbli...@ucsd.edu>
>>> wrote:
>>>
>>>> Thanks for the response, David!
>>>>
>>>> You can fetch biological assemblies from the PDB using `fetch PDBID,
>>>> type=pdb1` (or the PDB Loader Service plugin). This gives you the proper
>>>> oligomeric state, rather than just the asymmetric unit. However, since it
>>>> can potentially have chains from multiple asymmetric units, the chain IDs
>>>> are not necessarily unique. The RCSB's solution to this is to put each
>>>> asymmetric unit in a different state (arguably an abuse of the MODEL
>>>> record, but they write the spec).
>>>>
>>>> Intra_fit won't work for me because I want to align two different
>>>> objects. I can't use split_states because I want the whole multi-state
>>>> object to move together, but I want to only align based on a single
>>>> state/chain combo. Trying to do `align` on a multi-state object fails badly
>>>> because it aligns to the average position of equivalently-named atoms in
>>>> all states, which falls along a single line for symmetric structures. Thus
>>>> align basically just superimposes the centers of mass, even very similar
>>>> structures.
>>>>
>>>> I should also mention that I use most of the alignment methods
>>>> (pair_fit, align, super, cealign, etc), so fixing the select command or
>>>> adding selection language would be preferable to just fixing single
>>>> commands.
>>>>
>>>> -Spencer
>>>>
>>>> On Wed, Nov 5, 2014 at 1:03 PM, David Hall <li...@cowsandmilk.net>
>>>> wrote:
>>>>
>>>>> I'm a little confused by what PDB assemblies you are talking about
>>>>> that are distributed as multi-state structures, but assuming they are
>>>>> similar to NMR models... you can use intra_fit (
>>>>> http://www.pymolwiki.org/index.php/Intra_fit )
>>>>>
>>>>> For example, with 2kgx, we start out with chain a in state 1 and 2 and
>>>>> having an rms of 0.927 and chain b having an rms of 1.143
>>>>>
>>>>> intra_fit 2kgx and chain a
>>>>>
>>>>> now, chain a has an rms of 0.923 and chain b has an rms of 1.155
>>>>>
>>>>> intra_fit 2kgx and chain b
>>>>>
>>>>> now, chain a has an rms of 0.952 and chain b has an rms of 1.130
>>>>>
>>>>> Note, it can take a target state, that all other states are aligned
>>>>> to, etc.
>>>>>
>>>>>
>>>>> On Wed, Nov 5, 2014 at 5:23 AM, Spencer Bliven <sbli...@ucsd.edu>
>>>>> wrote:
>>>>>
>>>>>> Hey, I've been doing a lot of work with PDB assemblies, which are
>>>>>> distributed as multi-state structures. I would like to be able to align
>>>>>> them based on individual chains from individual states. However, I
>>>>>> haven't
>>>>>> been able to figure out any way to uniquely identify atoms by state.
>>>>>>
>>>>>> The select statement contains a `state` parameter, but as noted in this
>>>>>> 2010 thread
>>>>>> <https://www.mail-archive.com/pymol-users@lists.sourceforge.net/msg08224.html>
>>>>>> it doesn't seem to do anything. The thread also suggests using
>>>>>> cmd.get_model(), but I'm not sure how to get a selection based on a
>>>>>> chempy.models.Indexed instance.
>>>>>>
>>>>>> My current workaround is to split_states, use alter to make all the
>>>>>> chains unique, then merge the structures again. This is incredibly
>>>>>> tedious.
>>>>>> Does anyone know a better way?
>>>>>>
>>>>>> It would be nice if (1) the select command were fixed, and/or (2) add
>>>>>> a property selector for states.
>>>>>>
>>>>>> Cheers,
>>>>>> Spencer
>>>>>>
>>>>>>
>>>>>> ------------------------------------------------------------------------------
>>>>>>
>>>>>> _______________________________________________
>>>>>> PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net)
>>>>>> Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users
>>>>>> Archives:
>>>>>> http://www.mail-archive.com/pymol-users@lists.sourceforge.net
>>>>>>
>>>>>
>>>>>
>>>>
>>>
>>
>
------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=164703151&iu=/4140/ostg.clktrk
_______________________________________________
PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net)
Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users
Archives: http://www.mail-archive.com/pymol-users@lists.sourceforge.net
