[gmx-users] Fwd: PCA two ensembles different atoms order

[Nawel Mele]

HI Peter,

The correlation between the NMR and the simulation is pretty similar. I
have analysed torsion angles distributions and they do sample similar area.
When I performed the PCA I projected the NMR and REMD ensembles onto the
NMR PCs. I have four different compounds and for each of them I performed
REMD with two different starting conformations. I also have NMR ensemble
for all of them. After performing the PCA for all of them I noticed a
similar behaviour. First the NMR ensemble is spread almost evenly on the
two REMD ensembles (two starting confo) but the two REMD ensembles do not
overlap but are kind of separated into the PCs which made me wonder if it
was an issue of the files or if it is completely ok.



2017-07-03 12:40 GMT+01:00 Peter Kroon <p.c.kroon at rug.nl>:

> Hi Nawel,
>
>
> I'll admit that I'm out of my depth PCA-wise from here on out, so I
> heartily invite the rest of the list to chip in.
>
> Could it be that there's simply poor correlation between your
> simulations and your NMR ensemble?
>
>
> Peter
>
>
> On 03-07-17 12:34, Nawel Mele wrote:
> > Hi Peter,
> >
> > What I have is two REMD trajectories and an NMR ensemble. The two MD
> > trajectories were not written in the same way in term of atoms order. So
> I
> > just rewrite one of them to have the atoms in the pdb files in same order
> > in term of indices and same for the NMR ensemble so all three ensemble
> are
> > the same. For the reference structure I use the first frame of the first
> > REMD trajectories.
> >
> > I then did exactly what you said, I calculated the PC for one ensemble
> and
> > then then projected the two ensembles and the NMR ensemble onto those PCS
> > but the results obtained (the PC plots) was looking "weird".
> > So I decided to looked at my ensembles into VMD and I realised that if I
> > combined them in one big trajectory the bonds are kind of broken for the
> > trajectory that I rewrite (because I started the trjconv with the first
> > ensemble and VMD use the first frame of this ensemble as topology and it
> > looks like even tough the atoms are in same order vmd is having an
> issue).
> > I mean by that that if I open them separately it is fine but combined it
> is
> > like vmd does not recognise the atom order. I don't know if it affects
> the
> > PCA result.
> >
> > I don't know if it is clearer?
> >
> > Nawel
> >
> > 2017-07-03 10:44 GMT+01:00 Peter Kroon <p.c.kroon at rug.nl>:
> >
> >> Hi Nawel,
> >>
> >>
> >> I'm not quite sure what you're trying to achieve, or how, or what
> >> exactly is "wrong".
> >>
> >> In general though: calculate the principle components only once, for one
> >> of the ensembles, and project both ensembles on those PCs to get
> >> comparable results.
> >>
> >>
> >> Peter
> >>
> >>
> >> On 03-07-17 11:37, Nawel Mele wrote:
> >>> Hi all ,
> >>>
> >>> When performing a PCA of two different ensembles of the same molecule
> the
> >>> atoms of each molecule need to be in the same order ( same atom name
> >>> associated to the same index) right ?
> >>>
> >>> However, my two ensembles, even being the same molecule ( different
> >>> starting confirmation ), had their atoms in different order . So I
> simply
> >>> build a python script to rewrite the ensemble in a same way .
> Everything
> >>> went right but once I performed the PCA I obtain " wrong " result .
> >> Looking
> >>> at both ensembles separately in vmd they are fine , the rewrite
> ensemble
> >> is
> >>> not broken but if I combined the ensemble they don't " like each other
> "
> >>> like if they weren't the same molecule so don't have the same topology
> ,
> >>> I don't know if it clear .
> >>> Does anyone had to deal with that ?
> >>>
> >>> Regards,
> >>>
> >>> Nawel
> >>>
> >>>
> >>
> >>
> >> --
> >> Gromacs Users mailing list
> >>
> >> * Please search the archive at http://www.gromacs.org/Support
> >> /Mailing_Lists/GMX-Users_List before posting!
> >>
> >> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >>
> >> * For (un)subscribe requests visit
> >> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> >> send a mail to gmx-users-request at gromacs.org.
> >>
> >
> >
>
>
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at http://www.gromacs.org/
> Support/Mailing_Lists/GMX-Users_List before posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>



-- 

Nawel Mele, PhD Research Student

Jonathan W Essex Group, School of Chemistry

University of Southampton,  Highfield

Southampton, SO17 1BJ