On Sat, Mar 31, 2018 at 4:32 AM, Alexander Alexander
<[email protected]> wrote:
> Hi Christoph,
> Thank you again.
>
> On Saturday, March 31, 2018 at 5:02:49 AM UTC-4, Christoph Junghans wrote:
>>
>> On Fri, Mar 30, 2018 at 9:34 AM, Alexander Alexander
>> <[email protected]> wrote:
>> >
>> > Dear Christoph,
>> >
>> > Thank you for your informative response.
>> >
>> > From what I learned reading the manual and mailing list, below I
>> > summarize
>> > the step needed to parameterize the CG FF (both bonded and nonbonded)
>> > for a
>> > molecule using IBI, I am sure it save a lot of time for new users, I
>> > hope
>> > you could correct me wherever needed;
>> >
>> > What we have initially is single-mol-vacuum-aa.gro file.
>> >
>> > 1. Creating the desired mapping.xml from the single-mol-vacuum-aa.gro .
>> >
>> > 2. Generating the exclusions list using over the single molecule:
>> > csg_boltzmann --top single-mol-vacuum-aa.tpr --cg mapping.xml --excl
>> > exclusions.txt.
>> >
>> > 3. Adding the exclusions.txt to the single-mol-vacuum-aa.top and
>> > conduction
>> > an atomistic MD simulation for the single molecule in vacuum to have a
>> > new
>> > single-mol-vacuum-aa-excl.tpr and single-mol-vacuum-aa.trr .
>> >
>> > 4. Defining a setting-bonded.xml for the bonded interactions.
>> >
>> > 5. Calculating the bonded parameter using the "csg_stat --top
>> > single-mol-vacuum-aa-excl.tpr --trj single-mol-vacuum-aa.trr --cg
>> > mapping.xml --nt 4 --options setting-bonded.xml --begin 20000.
>> >
>> > Now we have all bonded interaction (bonded.dist.new renamed to
>> > bonded.dist.tgt) which can be taken later in IBI.
>> You could also Boltzmann invert them right now and only iterate the
>> non-bonded interactions to start with.
>>
> If the bonded parameters are also iterated, how much the computational cost
> increase usually?
If you update one interaction per iteration step, it is linear in cost.
But it is all about cross interaction correlation, stability and
complexity, start simple (non-bonded only).

> One more confirmation please concerning to the ensemble used in each
> simulation:
>  single-mole-vacume should be in NVT, should sd integrator is used here?
You don't have use sd, just make sure to sample in NVT, which could be
done with a thermostat as well.

> with md the final structure was normal whereas by sd integratorgave me a
> weird structre after 80 ns, something like the attached final gro file,
> although I used NPT, that might be the reason.
>
> bigger system atomistic simulation should be in NVT? Again, md or sd
> integrator? If so, I have to redo some simulation which I have performed in
> NPT with md integrator.
Again, most coarse-graining methods only work in NVT, so you should
sample atomistic and coarse-grained both in that ensemble.
If you do that with sd or a thermostat doesn't matter much.

>
> The iteration CG simulation of bonded and nonbonded should defiantly be
> performed in NVT with sd integrator.
>
> I used a <step>0.0001</step> in setting.xml, so that mmight be the reason of
> getting too noisy bonded.dist.new file specially for the angles, am I right
> please?
0.0001 seems a bit small, but that is really system dependent, just
remember that the smaller you make the bins the longer you need to run
to get enough statistics.

> Also, I see in the literature, like the original votca paper, that the
> angles have been plotted vs degree which is bettere, but what csg_stat gives
> me for angles is vs number not degree, I was wondering how I can convert
> that?
VOTCA does everything in radian, but you can just scale the whole
table if you need to.
However, for bonded IBI, this scaling will be done automatically.

>
> For a system with 5 bead if it is really necessary to define dihedral? what
> if I do not consider? In noticed that the dihedral has not been that much
> paid attention in your examples.
Again, depends on your system, start simple and ditch the dihedrals
for the start.

>>
>> >
>> > 6. Generating a bigger system from the single molecule,  bigger.gro and
>> > conducting an atomistic MD smilulation for the bigger system (again we
>> > have
>> > to use the exclusions.txt list here !? ).
>> Just do a normal full melt simulation (without exclusions) to get the
>> RDFs of the non-bonded interactions.
>
> Ok , thanks.
> But the exclusions should be considered in the IBI iteration by adding it to
> the topol.top file used in IBI iteration, if I am not wrong.!?
The exclusion are for the atomistic topology not the coarse-grained
one, which is used in IBI.

>>
>> >
>> > 7. Crreating a neww setting file in which only the nonbonded
>> > interactions
>> > have been defined, setting-nonbonded.xml.
>> >
>> > 8. Now we can calculate the RDF either using gmx rdf in gromacs or
>> > csg_stat
>> > in votca over the bigger system "csg_stat --top bigger.tpr --trj
>> > bigger.trr
>> > --cg mapping.xml --nt 4 --options setting-nonbonded.xml --begin 2000.
>> > The RDF is needed only for those pairs which we are interested in their
>> > nononbonded interaction.
>> >
>> > Now, we also have the nonbonded.dist.new renamed to nonbonded.dist.tgt
>> > files.
>> >
>> > 9. Combining the both settting-bonded.xml and setting-nonbonded.xml to a
>> > setting.xml and adding the parameters needed for IBI of both bonded and
>> > nonbonded to it.
>> >
>> > 10. Mapping the bigger atomistic system to a CG system to have
>> > bigger-CG.gro
>> > using "csg_map --top bigger.tpr --trj bigger.trr --cg mapping.xml --out
>> > CG.gro.
>> >
>> > 11. Having the bigger-CG.gro, setting.xml, bonded.dist.tgt and
>> > nonbonded.dist.tgt along with other files needed for a normal MD
>> > simulation
>> > using gromacs (coulombtype and vdw-type = User), we can start iteration
>> > (csg_inverse --options settings.xml) to have finally our
>> > table_bonded.xvg
>> > and table_nonbonded.xvg files.
>> >
>> >
>> > On Friday, March 30, 2018 at 4:51:53 AM UTC-4, Christoph Junghans wrote:
>> >>
>> >> On Thu, Mar 29, 2018 at 12:14 PM, Alexander Alexander
>> >> <[email protected]> wrote:
>> >> > Thank you.
>> >> >
>> >> > On Thursday, March 29, 2018 at 1:31:52 PM UTC-4, Christoph Junghans
>> >> > wrote:
>> >> >>
>> >> >> On Thu, Mar 29, 2018 at 10:25 AM, Alexander Alexander
>> >> >> <[email protected]> wrote:
>> >> >> >
>> >> >> > Thanks.
>> >> >> > I would be so thankful if one could comment on below questions;
>> >> >> >
>> >> >> > 1. The manual states that the IBI so far only supports iterative
>> >> >> > refinement
>> >> >> > of non-bonded interactions, however, I saw an example of
>> >> >> > setting.xml
>> >> >> > file in
>> >> >> > mailing list in which the bonded interactions were also defined
>> >> >> > for
>> >> >> > refinement by IBI, so, does IBI also support the bonded
>> >> >> > interactions?
>> >> >> Yes, it is supported for a while now!
>> >> >> It seems we forgot to update the manual, where in the manual is it
>> >> >> wrong? And can you send a pull request against
>> >> >> https://github.com/votca/csg-manual to fix it?
>> >> >
>> >
>> > The lines which need updating are in page 31 of manual section 7.2.1. I
>> > do
>> > not have github account to pull a request over there.
>> Fixed here: https://github.com/votca/csg-manual/pull/7
>> and that seems to be you: https://github.com/AlexanderWi (or somebody
>> is misusing your gmail address)
>
> That is me, but I am not aithub guy and do not use it, so I am not familiar
> with it. But I will try again.
>
> Best regards,
> Alex
>>
>>
>> Christoph
>> >
>> > Best regards,
>> > Alex
>> >>
>> >> >
>> >> > Yes I will.
>> >> >>
>> >> >>
>> >> >> >
>> >> >> > 2. I guess the ensemble chosen for MD simulation during IBI
>> >> >> > depends
>> >> >> > on
>> >> >> > the
>> >> >> > physics of the system (similar to the normal MD simulation),
>> >> >> > however,
>> >> >> > the
>> >> >> > Tcoupl = 0 and Pcoupl = 0 has been assigned for the temperature
>> >> >> > and
>> >> >> > pressure
>> >> >> > for all of the example of CG I have seen, so, I was wondering if
>> >> >> > the
>> >> >> > ensemble should be similar to the all atoms MD  simulation by
>> >> >> > which
>> >> >> > target_rdf.xvg has been calculated, am I right?
>> >> >> Yes, we usually do both simulations in NVT ensemble and we achieve
>> >> >> that with a stochastic dynamics (sd) integrator in Gromacs.
>> >> >>
>> >> >> >
>> >> >> > 3. We have to add the [ defaults ] and [ atomtypes ] section to
>> >> >> > the
>> >> >> > topol.top generated by csg_gmxtopol, if I am not wrong the  [
>> >> >> > defaults ]
>> >> >> > depends on the type of force field we want to use which should be
>> >> >> > again
>> >> >> > similar to the target all atom MD simulation, for the [ atomtypes
>> >> >> > ],
>> >> >> > how
>> >> >> > I
>> >> >> > should indicate the C6 and C12 parameter for each bead? of course
>> >> >> > the
>> >> >> > C6
>> >> >> > and
>> >> >> > C12 are available in ffnonbonded.itp for individual atoms involved
>> >> >> > in
>> >> >> > each
>> >> >> > beads but not for the beads itself. For instance to get the mass
>> >> >> > of
>> >> >> > bead
>> >> >> > we
>> >> >> > sum up the mass of individual atoms involved in that bead.
>> >> >> When VOTCA creates the xvg tables for Gromacs, it writes the
>> >> >> potential
>> >> >> and forces in the C12 column, so you should set C12 to 1.0 (except
>> >> >> if
>> >> >> you need to scale it for some reason) and C6 to 0.0 (or whatever).
>> >> >>
>> >> > If I simply copy the content of exclusions.txt to my topol.top file,
>> >> > then
>> >> > everything will be fine an no mismatching shows up between the rdf's
>> >> > (if
>> >> > rest is fine)?
>> >> The exclusions are for the atomistic vacuum run, but yes, the theory
>> >> is that the exclusions will help.
>> >>
>> >> >>
>> >> >> >
>> >> >> > 4. Below is a part of setting.xml file used in a IBI simulation;
>> >> >> >
>> >> >> >      <name>ABA</name>
>> >> >> >     <min>1.2</min>
>> >> >> >     <max>2.9</max>
>> >> >> >     <step>0.001</step>
>> >> >> >     <inverse>
>> >> >> >       <!-- target distribution -->
>> >> >> >       <target>angle1.dist.tgt</target>
>> >> >> >       <do_potential>0 0 0</do_potential>
>> >> >> >       <post_update>smooth</post_update>
>> >> >> >       <post_update_options>
>> >> >> >         <scale>0.001</scale>
>> >> >> >       </post_update_options>
>> >> >> >       <post_add>convergence</post_add>
>> >> >> >       <gromacs>
>> >> >> >         <table>table_a1.xvg</table>
>> >> >> >       </gromacs>
>> >> >> >     </inverse>
>> >> >> >   </bonded>
>> >> >> >   <bonded>
>> >> >> > Can you please confirm me that the
>> >> >> > <target>angle1.dist.tgt</target>
>> >> >> > is
>> >> >> > the
>> >> >> > one we provide in advance and the  <table>table_a1.xvg</table> is
>> >> >> > the
>> >> >> > one
>> >> >> > which votca results after refinement? (in general for refinement
>> >> >> > of
>> >> >> > bonded
>> >> >> > and nonbonded interaction <target>case.tgt</target> is the one we
>> >> >> > provide
>> >> >> > and  <table>case.table.xvg</table> is the votca give finally)
>> >> >> > Also, how I can calculate the "angle1.dist.tgt" one?
>> >> >> yes, <target> is the distribution VOTCA compares the current
>> >> >> distribution against and <table> is the table VOTCA will generate
>> >> >> from
>> >> >> the update of that comparison, the filename is listed, so that you
>> >> >> can
>> >> >> setup your topology to use exactly the table with the same number.
>> >> >>
>> >> > Is the angle1.dist.tgt or bond.dist.tgt are the ones that
>> >> > "csg_boltsman
>> >> > hist
>> >> > ..." gives us for a single molecule in vacuum? Any further refinement
>> >> > is
>> >> > needed to be performed on what hist gives us?
>> >> Yes, csg_boltzmann or csg_stat, which can also calculate bonded
>> >> distributions as well.
>> >> Refinement is only needed, if you have some invalid point in between
>> >> min and max, otherwise you should be good.
>> >> For IBI of bonded interactions, you should use the same normalization
>> >> as csg_stat does, i.e. calculated the distributions with csg_stat,
>> >> e.g. with:
>> >> $ csg_stat --trj atomistic.xtc --top atomistic.tpr --cg mapping.xml
>> >> --options interactions.xml
>> >>
>> >> >
>> >> > In the hexane ibi_all example, there is a table.xvg file in the
>> >> > parent
>> >> > folder, I was wondering if we have to provide it in advance ourselves
>> >> > or
>> >> > the
>> >> > the votca would calculate it in the first step_000?
>> >> The table.xvg was needed for older versions of gromacs and if I recall
>> >> correctly not needed anymore with gromacs-2016.
>> >> In a nutshell, gromacs needed a catch-all table for all the
>> >> interactions not covered by special interactions e.g. table_CG_CG.xvg,
>> >> even if the rest group was empty.
>> >> If all interactions are covered by special interactions, which is
>> >> usually the case, then table.xvg could be a table full of zeros.
>> >>
>> >> Christoph
>> >> >
>> >> > Thank you.
>> >> >
>> >> > Best regards,
>> >> > Alex
>> >> >
>> >> >>
>> >> >> >
>> >> >> > How should we know the <min>xx</min> and <max>xxx</max> for each
>> >> >> > interaction?
>> >> >> Just look at the distributions and pick min and max so that the
>> >> >> distribution is bigger than 0 and that the extrapolation of the
>> >> >> potentials will work (which is a bit of an art in itself).
>> >> >>
>> >> >> Christoph
>> >> >>
>> >> >> >
>> >> >> > Thank you very much in advance for your time.
>> >> >> >
>> >> >> > Best regards,
>> >> >> > Alex
>> >> >> >
>> >> >> > On Wednesday, March 28, 2018 at 3:28:44 AM UTC-4, Christoph
>> >> >> > Junghans
>> >> >> > wrote:
>> >> >> >>
>> >> >> >> On Tue, Mar 27, 2018 at 6:28 PM, Alexander Alexander
>> >> >> >> <[email protected]> wrote:
>> >> >> >> >
>> >> >> >> > Thank Christoph.
>> >> >> >> > Actually the xmlwf did not give more information than the parse
>> >> >> >> > error
>> >> >> >> > in
>> >> >> >> > the
>> >> >> >> > last line.
>> >> >> >> >
>> >> >> >> > I will take the opportunity to ask one more general question:
>> >> >> >> >
>> >> >> >> > Using the boltzmann inversion and iterative boltzmann inversion
>> >> >> >> > I
>> >> >> >> > wand
>> >> >> >> > to
>> >> >> >> > obtain the coarse-grained parameters (bonded and nonbonded) for
>> >> >> >> > a
>> >> >> >> > molecule,
>> >> >> >> > the molecule has 53 atom which I have mapped them on 5 beads (3
>> >> >> >> > types)
>> >> >> >> > and 4
>> >> >> >> > bonds and 3 angles. The all atom MD simulation for a system
>> >> >> >> > containing
>> >> >> >> > 307
>> >> >> >> > molecules of my interest using gromacs and gromos force filed
>> >> >> >> > has
>> >> >> >> > also
>> >> >> >> > been
>> >> >> >> > completed.
>> >> >> >> > As a new user, I found the votca manual too scattered (sorry
>> >> >> >> > for
>> >> >> >> > saying
>> >> >> >> > that) to follow in order to obtain the CG parameters, I tried
>> >> >> >> > to
>> >> >> >> > understand
>> >> >> >> > the examples but still I am not able to fulfill the procedure.
>> >> >> >> >
>> >> >> >> > By considering the explained molecule I would be so appreciated
>> >> >> >> > if
>> >> >> >> > one
>> >> >> >> > could
>> >> >> >> > kindly itemize the steps I have to take to obtain all of the CG
>> >> >> >> > parameters
>> >> >> >> > of the molecule of my interest.
>> >> >> >> There is no general practice, you will need to check the
>> >> >> >> literature
>> >> >> >> and find a system close to what you want to coarse-grain.
>> >> >> >> You could trying to coarse-grain all interactions at once, but
>> >> >> >> usually
>> >> >> >> something will not work and make the system explode.
>> >> >> >> I prefer to start small and build it up, so a.) Boltzmann invert
>> >> >> >> the
>> >> >> >> bonded interactions from a single molecule (in vacuum)
>> >> >> >> trajectory,
>> >> >> >> then b.) the non-bonded interactions with iterative Boltzmann
>> >> >> >> inversion and last you might want to c.) re-iterate the bonded
>> >> >> >> interactions.
>> >> >> >>
>> >> >> >> This has already been extensively discuss on this mailing list,
>> >> >> >> so
>> >> >> >> have a look in the archive. The hexane tutorial might be worth
>> >> >> >> looking
>> >> >> >> at, too.
>> >> >> >>
>> >> >> >> Christoph
>> >> >> >> >
>> >> >> >> > Thank you very much.
>> >> >> >> >
>> >> >> >> > Best regards,
>> >> >> >> > Alex
>> >> >> >> >
>> >> >> >> >
>> >> >> >> >
>> >> >> >> >
>> >> >> >> > On Tuesday, March 27, 2018 at 4:55:51 PM UTC-4, Christoph
>> >> >> >> > Junghans
>> >> >> >> > wrote:
>> >> >> >> >>
>> >> >> >> >> On Tue, Mar 27, 2018 at 2:07 PM, Alexander Alexander
>> >> >> >> >> <[email protected]> wrote:
>> >> >> >> >> >
>> >> >> >> >> > Solved now, I was missing a backslash in </ident>.
>> >> >> >> >> > Anyway, thank you.
>> >> >> >> >> The expat package, provides a little tool called "xmlwf",
>> >> >> >> >> which
>> >> >> >> >> allows
>> >> >> >> >> you to check whether an xml file is well-formatted.
>> >> >> >> >>
>> >> >> >> >> Christoph
>> >> >> >> >> >
>> >> >> >> >> > Regards,
>> >> >> >> >> > Alex
>> >> >> >> >> >
>> >> >> >> >> >
>> >> >> >> >> >
>> >> >> >> >> > On Tuesday, March 27, 2018 at 2:20:10 PM UTC-4, Alexander
>> >> >> >> >> > Alexander
>> >> >> >> >> > wrote:
>> >> >> >> >> >>
>> >> >> >> >> >>
>> >> >> >> >> >> Dear all,
>> >> >> >> >> >>
>> >> >> >> >> >> The "csg_dump --top case.tpr --cg case.xml" parses error at
>> >> >> >> >> >> the
>> >> >> >> >> >> last
>> >> >> >> >> >> line
>> >> >> >> >> >> of my case.xml with the mismatched tag, I can not find out
>> >> >> >> >> >> where
>> >> >> >> >> >> I
>> >> >> >> >> >> am
>> >> >> >> >> >> doing
>> >> >> >> >> >> wrong, so, I would be so appreciated if you could have a
>> >> >> >> >> >> look
>> >> >> >> >> >> at
>> >> >> >> >> >> my
>> >> >> >> >> >> attached
>> >> >> >> >> >> case.xml,.top,gro files and help me overcome the issue.
>> >> >> >> >> >>
>> >> >> >> >> >> Thanks
>> >> >> >> >> >> Best regards,
>> >> >> >> >> >> Alex
>> >> >> >> >> >
>> >> >> >> >> > --
>> >> >> >> >> > You received this message because you are subscribed to the
>> >> >> >> >> > Google
>> >> >> >> >> > Groups
>> >> >> >> >> > "votca" group.
>> >> >> >> >> > To unsubscribe from this group and stop receiving emails
>> >> >> >> >> > from
>> >> >> >> >> > it,
>> >> >> >> >> > send
>> >> >> >> >> > an
>> >> >> >> >> > email to [email protected].
>> >> >> >> >> > To post to this group, send email to [email protected].
>> >> >> >> >> > Visit this group at https://groups.google.com/group/votca.
>> >> >> >> >> > For more options, visit https://groups.google.com/d/optout.
>> >> >> >> >>
>> >> >> >> >>
>> >> >> >> >>
>> >> >> >> >> --
>> >> >> >> >> Christoph Junghans
>> >> >> >> >> Web: http://www.compphys.de
>> >> >> >> >
>> >> >> >> > --
>> >> >> >> > You received this message because you are subscribed to the
>> >> >> >> > Google
>> >> >> >> > Groups
>> >> >> >> > "votca" group.
>> >> >> >> > To unsubscribe from this group and stop receiving emails from
>> >> >> >> > it,
>> >> >> >> > send
>> >> >> >> > an
>> >> >> >> > email to [email protected].
>> >> >> >> > To post to this group, send email to [email protected].
>> >> >> >> > Visit this group at https://groups.google.com/group/votca.
>> >> >> >> > For more options, visit https://groups.google.com/d/optout.
>> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >> >> --
>> >> >> >> Christoph Junghans
>> >> >> >> Web: http://www.compphys.de
>> >> >> >
>> >> >> > --
>> >> >> > You received this message because you are subscribed to the Google
>> >> >> > Groups
>> >> >> > "votca" group.
>> >> >> > To unsubscribe from this group and stop receiving emails from it,
>> >> >> > send
>> >> >> > an
>> >> >> > email to [email protected].
>> >> >> > To post to this group, send email to [email protected].
>> >> >> > Visit this group at https://groups.google.com/group/votca.
>> >> >> > For more options, visit https://groups.google.com/d/optout.
>> >> >>
>> >> >>
>> >> >>
>> >> >> --
>> >> >> Christoph Junghans
>> >> >> Web: http://www.compphys.de
>> >> >
>> >> > --
>> >> > You received this message because you are subscribed to the Google
>> >> > Groups
>> >> > "votca" group.
>> >> > To unsubscribe from this group and stop receiving emails from it,
>> >> > send
>> >> > an
>> >> > email to [email protected].
>> >> > To post to this group, send email to [email protected].
>> >> > Visit this group at https://groups.google.com/group/votca.
>> >> > For more options, visit https://groups.google.com/d/optout.
>> >>
>> >>
>> >>
>> >> --
>> >> Christoph Junghans
>> >> Web: http://www.compphys.de
>> >
>> > --
>> > You received this message because you are subscribed to the Google
>> > Groups
>> > "votca" group.
>> > To unsubscribe from this group and stop receiving emails from it, send
>> > an
>> > email to [email protected].
>> > To post to this group, send email to [email protected].
>> > Visit this group at https://groups.google.com/group/votca.
>> > For more options, visit https://groups.google.com/d/optout.
>>
>>
>>
>> --
>> Christoph Junghans
>> Web: http://www.compphys.de
>
> --
> You received this message because you are subscribed to the Google Groups
> "votca" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> To post to this group, send email to [email protected].
> Visit this group at https://groups.google.com/group/votca.
> For more options, visit https://groups.google.com/d/optout.



-- 
Christoph Junghans
Web: http://www.compphys.de

-- 
You received this message because you are subscribed to the Google Groups 
"votca" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/votca.
For more options, visit https://groups.google.com/d/optout.

Reply via email to