Hi Yikan,
I'd like to make a small clarification, which, although you may not need need, might benefit others reading this. The script described in our recent publication deals with protein structure calculation. Although it should be definitely helpful to understand scripts dealing with other systems, it obviously differs slightly from scripts used in RNA structure calculation. For example, the HBDB term addresses H-bonding in protein but not nucleic acids. The equivalent script(s) to deal with RNA are found in the eginput/rna directory (within the Xplor-NIH installation directory). The script I sent you earlier for you RNA problem was based on that. Best, Guillermo ________________________________ From: ZhangYikan <[email protected]> Sent: Wednesday, December 13, 2017 11:56 AM To: Bermejo, Guillermo (NIH/CIT) [E]; [email protected] Subject: Re: [Xplor-nih] Wrong topology structures in conformational sampling hi Guillermo, Thank you for your sincerely help! Now, I am trying to find an appropriate repel value. But I have two questions about Xplor-NIH: 1) About random seed:"the same seed number on the same computer yields the same results” In oder to generate more structures, I have submitted many jobs to the cluster. like “ #!/bin/sh #8st job k=8 cd /Share/home/fangxy/Users/Zhangyk/Zxplor-test/Gwnvpks/sec_wp/secout/8 xplor -py ../../wp.py ../../${k}.pdb” in these jobs, the random seeds in the “wp.py” are all the same, so I do not know whether this will effect the results of the conformational sampling. 2) “https://link.springer.com/protocol/10.1007/978-1-4939-7386-6_14” mentioned that “def calcOneStructure(loopInfo):” is consist of “High Temperature Stage” “Simulated Annealing Stage” “Torsion Angle Minimization” and so on. One simulate annealing stage(temp from 3500 to 25) will calculate only one structure. And when I checked the output pdb, I found that the two outputs of adjacent numbers(like 1.pdb, 2.pdb) only have a little similarity and that is OK for the sampling. So if we want to sampling more conformations, do we have some other parameter to adjust (in the three stages) ? And do the numberOfStructures has "upper limit” ? Thank you! Best, Yikan Ph. D candidate TsingHua University School of life Sciences Beijing, China tel:010-62773783 在 2017年12月13日，上午4:47，Bermejo, Guillermo (NIH/CIT) [E] <[email protected]<mailto:[email protected]>> 写道： Hi Yikan, The nonbonded potential that we generally use (and you're using now) is purely repulsive. To account for the lack of an attractive component and avoid having expanded structures, atomic radii are scaled down by a factor called "repel". For the current default parameters (which you are implicitly using by running the function protocol.loadPDB), we found the optimal repel value to be 0.9. In general, one doesn't explicitly deal with repel, as the function that sets up the potential automatically handle it (they even recognize the version of the parameters used, for which there may be different optimal repel values). (An exception is during the high temperature stage, where we use expanded atoms with repel=1.2 to improve sampling; see the highTempParams setup.) In your particular case, you want to actively change repel by replacing the following line rampedParams.append( StaticRamp("initRepel(repel,use14=False)") ) by rampedParams.append( MultRamp(0.9, 1.0, "initRepel(repel, use14=False, repel=VALUE)") ) where repel will be increased from 0.9 to 1 during simulated annealing. You can play with different repel values. Note that there's another "repel" word here that refers to the instance of the repulsive potential [repel = create_RepelPot('repel')], and which could, in principle, be given any arbitrary name. The way of varying repel is general, and is applied to other things (e.g., force constants). For more details, see a recent publication where we explain a structure calculation script (https://link.springer.com/protocol/10.1007/978-1-4939-7386-6_14). Best, Guillermo ________________________________ From: ZhangYikan <[email protected]<mailto:[email protected]>> Sent: Tuesday, December 12, 2017 11:08 AM To: Bermejo, Guillermo (NIH/CIT) [E]; [email protected]<mailto:[email protected]> Subject: Re: [Xplor-nih] Wrong topology structures in conformational sampling hi Guillermo, Thank you for your help! in your previous mail you said “to increase the atomic radii (see the 'repel' argument in initRepel)”. But in the script “from repelPotTools import create_RepelPot,initRepel repel = create_RepelPot('repel') potList.append(repel) rampedParams.append( StaticRamp("initRepel(repel,use14=False)") ) rampedParams.append( MultRamp(.004, 4, "repel.setScale( VALUE)") ) # nonbonded interaction only between C1' atoms highTempParams.append( StaticRamp("""initRepel(repel, use14=True, scale=0.004, repel=1.2, moveTol=45, interactingAtoms="name P" )""") ) “ I can not find the terms that stands for “atom radius” and how to “increase the atomic radii(in intRepel)”. ps: Only RepelPot has the similar term “rMin”; but when I have looked through the examples in eginput, I did not find the similar use of “RepPot”. Thank you ! Best, Yikan Ph. D candidate TsingHua University School of life Sciences Beijing, China tel:010-62773783 在 2017年12月12日，下午11:38，Bermejo, Guillermo (NIH/CIT) [E] <[email protected]<mailto:[email protected]>> 写道： Hi Yikan, I think using Monte Carlo (in the torsion angle randomization) is appropriate. Good luck, Guillermo ________________________________ From: ZhangYikan <[email protected]<mailto:[email protected]>> Sent: Monday, December 11, 2017 6:36 PM To: Bermejo, Guillermo (NIH/CIT) [E]; [email protected]<mailto:[email protected]> Subject: Re: [Xplor-nih] Wrong topology structures in conformational sampling hell Guillermo, By the way, although I have known that the comformational sampling problem has been an untrival problem util now; how can I make sure that I have sampled almost all possible conformations of this RNA? Yikan Ph. D candidate TsingHua University School of life Sciences Beijing, China tel:010-62773783 > 在 2017年12月12日，上午7:25，ZhangYikan > <[email protected]<mailto:[email protected]>> 写道： > > hello Guillermo, > Thank you for your scripts, I have tried it. I found that there are still > some wrong topology structures and clahes(some overlaps):1) I will try to fix > the repel parameter and rerun the script; > 2) in the outputs of the previous zp_0-3.py, I found the terms of “impr” > seems unusual(in the .viols), did you have some tools to analysis this energy > term ? > > Thank you! > Best, > > Yikan > Ph. D candidate > TsingHua University > School of life Sciences > Beijing, China > tel:010-62773783 > > > >> 在 2017年12月12日，上午2:00，Bermejo, Guillermo (NIH/CIT) [E] >> <[email protected]<mailto:[email protected]>> 写道： >> >> Hi Yikan, >> >> I notice that your script seems to be more complicated than it has to be. >> I'm not sure this will solve your problem, but I'm attaching a simpler >> script based on eginput/rna/fold.py. It's basically the script used to >> fold an RNA molecule from an extended conformation (described in the paper I >> mentioned before). I modified it so that you load your input pdb structure >> and define the different rigid bodies (look for the lines with an inline >> comment "# COMPLETE !", where you must enter the right information). Then, >> it does torsion angle dynamics using the remaining torsional degrees of >> freedom (i.e., at the linker(s)). >> >> I commented all the experimental restraints. If you want to have the radius >> of gyration term (which appears in your script), you can introduce it >> yourself. Also, you might want to use it as a starting point for more >> complicated stuff, like "breaking" the linkers and randomizing the position >> of the domains. >> >> A suggestion to try to avoid (and detect) knotted structures is to increase >> the atomic radii (see the 'repel' argument in initRepel). >> >> Best, >> >> Guillermo >> >> From: ZhangYikan <[email protected]<mailto:[email protected]>> >> Sent: Monday, December 11, 2017 10:43 AM >> To: Bermejo, Guillermo (NIH/CIT) [E]; >> [email protected]<mailto:[email protected]> >> Subject: Re: [Xplor-nih] Wrong topology structures in conformational sampling >> >> hello Guillermo, >> >> I want these domains move (as rigid body)randomly and freely to sample all >> possible conformations of the RNA. >> >> Yikan >> Ph. D candidate >> TsingHua University >> School of life Sciences >> Beijing, China >> tel:010-62773783 >> >> >> >>> 在 2017年12月11日，下午11:27，Bermejo, Guillermo (NIH/CIT) [E] >>> <[email protected]<mailto:[email protected]>> 写道： >>> >>> Hi Yikan, >>> >>> Can you tell me what you are trying to do? >>> >>> Best, >>> >>> Guillermo >>> >>> >>> From: ZhangYikan <[email protected]<mailto:[email protected]>> >>> Sent: Friday, December 8, 2017 6:56 PM >>> To: Bermejo, Guillermo (NIH/CIT) [E]; >>> [email protected]<mailto:[email protected]> >>> Subject: Re: [Xplor-nih] Wrong topology structures in conformational >>> sampling >>> >>> hi Guillermo, >>> yes, i have received your last mails. And the attachment is my scripts. >>> >>> >>> Yikan >>> Ph. D candidate >>> TsingHua University >>> School of life Sciences >>> Beijing, China >>> tel:010-62773783 >>> >>> >>> >>>> 在 2017年12月9日，上午12:43，Bermejo, Guillermo (NIH/CIT) [E] >>>> <[email protected]<mailto:[email protected]>> 写道： >>>> >>>> Hi Yikan, >>>> >>>> I realized that my original response was only sent to you and not included >>>> in the mailing list. Here it is. >>>> >>>> One possibility is that there's something wrong with your script(s). >>>> >>>> I recommend basing your calculations on the scripts found in eginput/rna >>>> directory (within your Xplor-NIH directory). There you'll find: >>>> >>>> * fold.py, for initial calculations starting from an extended conformation. >>>> * refine.py, for subsequent refinement. >>>> >>>> In addition, you'll find all the restraints for an example system, in case >>>> you want to play with the scripts before venturing into your data. The >>>> scripts are highly commented, which should help in customizing them for >>>> your needs. There's also a paper describing them [Bermejo et al., 2016, >>>> Structure 24, 806-815]. >>>> >>>> Best, >>>> >>>> Guillermo >>>> >>>> >>>> >>>> >>>> >>>> From: >>>> [email protected]<mailto:[email protected]> >>>> >>>> <[email protected]<mailto:[email protected]>> >>>> on behalf of ZhangYikan >>>> <[email protected]<mailto:[email protected]>> >>>> Sent: Friday, December 8, 2017 6:53 AM >>>> To: [email protected]<mailto:[email protected]> >>>> Subject: [Xplor-nih] Wrong topology structures in conformational sampling >>>> >>>> >>>> >>>> hello all, >>>> >>>> Recently, I am doing conformational sampling with Xplor-NIH. And I got >>>> many pdbs, but when I check them I found that some pdbs have wrong >>>> topology structures like the attachments(in the pink cycle ): The RNA >>>> linker did not run with the right way and It went through the helix. I >>>> have tried to pick the structures by the “total energy” or “repel” terms, >>>> but it make no sense. Are there someone know how to kick out these wrong >>>> structures or to void these errors in the sampling process? Thank you! >>>> Yikan >>>> Ph. D candidate >>>> TsingHua University >>>> School of life Sciences >>>> Beijing, China >>>> tel:010-62773783 >> >> <fold.py> > > _______________________________________________ > Xplor-nih mailing list > [email protected]<mailto:[email protected]> > https://dcb.cit.nih.gov/mailman/listinfo/xplor-nih Xplor-nih Info Page<https://dcb.cit.nih.gov/mailman/listinfo/xplor-nih> dcb.cit.nih.gov<http://dcb.cit.nih.gov/> To see the collection of prior postings to the list, visit the Xplor-nih Archives. Using Xplor-nih: To post a message to all the list members ...
_______________________________________________ Xplor-nih mailing list [email protected] https://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
