[chimera-dev] Summit supercompuer

Eric Pettersen pett at cgl.ucsf.edu
Tue Jan 22 11:42:12 PST 2019


Hi Victor,
	So you want to run Chimera on a supercomputer because Chimera is such a memory hog. :-(
	As Elaine mentioned, we don’t have the manpower to devote major effort to both Chimera and ChimeraX unfortunately.  So we could not get directly involved in porting Chimera to Summit, though if you were to try it yourself we would be willing to try to any questions or issues you have on this mailing list.
	Nonetheless, I feel you would be far better served to use ChimeraX for this.  One of it’s major design goals was performance and efficiency.  For example with 5vf3 (Bacteriophage T4 isometric capsid protein), ChimeraX opens it in a second and displays the full capsid (60 copies) in under 3 seconds.  Chimera opens it in about 2.5 seconds and on my machine never displays the full capsid, taking a few minutes to crash after consuming over 14GB of memory.  Also, if there are features you need that are missing, there is a chance they would be implemented fairly quickly upon request (depends on difficulty, general usefulness, other priorities, etc.).

—Eric

> On Jan 22, 2019, at 11:02 AM, Elaine Meng <meng at cgl.ucsf.edu> wrote:
> 
> I don’t presume to answer this myself, but I expect the answer will be “No, we are now working on ChimeraX and one of its major advantages is increased ability to work with very large structures.”
> 
> Also yes, ChimeraX does have multiscale models low-res surfaces, just not the Multiscale Models GUI.
> 
> Symmetry copies as low-res surfaces only are specified with the “surfaceOnly true” option of the “sym” command in ChimeraX.
> 
> <http://rbvi.ucsf.edu/chimerax/docs/user/commands/sym.html>
> <http://rbvi.ucsf.edu/chimerax/docs/user/commands/sym.html#copies>
> 
>> On Jan 18, 2019, at 5:33 PM, Victor Padilla-Sanchez <70padillasan at cua.edu> wrote:
>> 
>> attachments
>> 
>> On Fri, Jan 18, 2019 at 8:27 PM Victor Padilla-Sanchez <70padillasan at cua.edu> wrote:
>> Hi Eric,
>> 
>> Thank you very much for your answer.
>> I need Chimera in Summit because I want to use their big ram memory to be able to visualize huge protein assemblies like the entire bacteriophage t4 virus and beyond and only a supercomputer can achieve that kind of memory.
>> It wouldn't be important if molecular surfaces can't be processed.
>> 
>> What about higher order structures like multi scale models ?
>> Since I work a lot with them.
>> 
>> I can see that Chimera X does not have that option.
>> 
>> The Director Jack Wells at Oak Ridge National Laboratory wants to know if you can be involved in being successfully porting Chimera in Summit.
>> Actually I asked him for permission to install Chimera but the engineer Jack Morrison saw the disclaimer and said is not possible.
>> I am trying to scale up my work with Chimera to a higher level but I need a supercomputer to do that.
>> I am using Rhea in Oak Ridge but it will be decommission therefore I want to try Summit if possible.
>> 
>> Would you be able to help us installing Chimera in Summit ?
>> 
>> Please let me know if you need further explanations of this work.
>> 
>> I am sending an attachment figure which I am able to do in their big supercomputer Rhea and which I want to continue in Summit.
>> They use power pc so the engineer said is not possible.
>> But apparently is possible. It is just that will take more effort if I understand correctly.
>> 
>> Please help me install Chimera in Summit. I think I will be able to achieve a lot in Summit.
>> 
>> Kind regards,
>> 
>> Victor Padilla-Sanchez, PhD
>> 
>> 
>> On Fri, Jan 18, 2019 at 7:51 PM Eric Pettersen <pett at cgl.ucsf.edu> wrote:
>> Hi Victor,
>> 	You haven’t said anything about your motivation for porting Chimera to Summit, so it’s hard to provide good guidance here.  That disclaimer about the code means that you can in fact compile Chimera from the provided source, but the result will not be able to compute/show molecular surfaces or any attributes based on such calculations (such as solvent-exposed surface area).  Depending on why you need Chimera on Summit, this may or may not be good enough.  The other disclaimer in the code about getting the compile to work being challenging is absolutely true, so that should be taken into account when deciding what course of action you want to take from here.
>> 	Another option to consider is compiling ChimeraX on Summit instead.  Again, since I don’t know your motivation I can’t say if ChimeraX’s current capabilities would meet your needs, but it seemed like something worth mentioning.
>> 
>> —Eric
>> 
>>> On Jan 18, 2019, at 8:31 AM, Victor Padilla-Sanchez <70padillasan at cua.edu> wrote:
>>> 
>>> Hi Eric,
>>> 
>>> The engineer at Oak Ridge says Summit architecture is incompatible:
>>> 
>>> SummitDev is based on the PowerPC (ppc64le) architecture, and is binary
>>>    incompatible with Chimera as it's distributed. The Chimera
>>>    documentation warns
>>>    that "it's not possible to re-compile a fully functional Chimera from
>>>    the
>>>    source code distributed here."
>>> 
>>> Do you know anyway around about this problem so we are able to install Chimera in Summit at Oak Ridge ?
>>> 
>>> Please let me know
>>> 
>>> Regards,
>>> 
>>> Victor
>>> 
>>> On Thu, Jan 10, 2019 at 1:34 PM Eric Pettersen <pett at cgl.ucsf.edu> wrote:
>>> Hi Victor,
>>> 	As far as we know, no one has specifically ported Chimera to Oak Ridge’s Summit.  I’m not sure why you would need Chimera to run on a supercomputer, but nonetheless Summit runs Red Had Enterprise 7.4, so I see no reason that the “headless” version of Chimera wouldn't just work.
>>> 
>>> —Eric
>>> 
>>> 	Eric Pettersen
>>> 	UCSF Computer Graphics Lab
>>> 
>>> 
>>>> On Jan 10, 2019, at 3:53 AM, Victor Padilla-Sanchez <70padillasan at cua.edu> wrote:
>>>> 
>>>> Dear Chimera,
>>>> 
>>>> Would you tell me if someone has developed/ported Chimera to Summit supercomputer at Oak Ridge National Laboratory ?
>>>> 
>>>> Please let me know,
>>>> 
>>>> Thank you very much,
>>>> 
>>>> Victor Padilla-Sanchez, PhD
>>>> _______________________________________________
>>>> Chimera-dev mailing list
>>>> Chimera-dev at cgl.ucsf.edu
>>>> http://plato.cgl.ucsf.edu/mailman/listinfo/chimera-dev
>>> 
>>> _______________________________________________
>>> Chimera-dev mailing list
>>> Chimera-dev at cgl.ucsf.edu
>>> http://plato.cgl.ucsf.edu/mailman/listinfo/chimera-dev
>> 
>> <2019-01-16.png><2019-01-18.png>_______________________________________________
>> Chimera-dev mailing list
>> Chimera-dev at cgl.ucsf.edu
>> http://plato.cgl.ucsf.edu/mailman/listinfo/chimera-dev
> 
> 
> _______________________________________________
> Chimera-dev mailing list
> Chimera-dev at cgl.ucsf.edu
> http://plato.cgl.ucsf.edu/mailman/listinfo/chimera-dev
> 




More information about the Chimera-dev mailing list