Dear Davide, Using one CG force field for the entire set of conformations of a protein can get quite tricky. There's two issues here: (1) CG resolution and (2) sampling.
(1) CG resolution: there's quite a history of CG models that have tried folding peptide/proteins (not necessarily systematically derived, many of them used a top-down approach). The general consensus is that it's very difficult to describe both alpha-helices and beta-sheets if, say, you use only one CG bead per residue backbone. For instance, MARTINI requires different force field for the two folds. Other models that have higher resolutions on the backbone can describe both folds, at least for simple peptides (see papers from Mousseau and Derreumaux, Mohanty and Irbäck, myself and Deserno). (2) When systematically CGing a peptide/protein, two main routes have been undertaken: - sample atomistically the interaction of pairs of amino acids in water. I remember a couple of papers from Marcos Betancourt on that subject a few years ago. The obvious issue here is that you don't get any correlations in your CG potentials. - sample atomistically an entire peptide/protein. That's what the two papers Christoph mentions are about. Note that adequate atomistic sampling becomes a critical issue very quickly--you need *a lot* of sampling. That's because protein interactions are so weakly stabilized. The way I see it is that you'll already need a microsecond of atomistic simulation for very small peptides--so that quickly gets out of hands. Transferability issues (i.e., sampling one protein and using the CG potentials on another) are another concern. Ideally, you'd want to have many proteins in your parametrization set (like in the paper of Ronald Hills). Generally, the scarcity of publications on the topic you describe is not due to people's lack of motivation or interest (it's actually quite exciting), but rather that it's a difficult problem. Best, Tristan On Tue, Mar 26, 2013 at 12:30 AM, Christoph Junghans <[email protected]>wrote: > 2013/3/25 Mercadante, Davide <[email protected]>: > > Dear Christoph, > > > > Many thanks for your reply that I have very much appreciated. > > > > No, nothing so top secrets. I am just trying to reproduce the full atom > > Potential Mean Force (PMF) of a protein in a CG fashion so that we could > > scale up the simulation in terms of system size. Other coarse grained ff > > have failed reproducing the Full atom PMF and it looks that votca can > > finally do the job..however, one of my question is if I would have a > > potential that is conformation-dependent. For example: if I have two > > conformations 1) extended 2) collapsed of the same peptide, will I have > > two different (or very different) potentials (reflected in the > topologies) > > because my full atom simulations were performed on different > > conformational ensembles? I hope not as the forcefield should be > > conformation-independent right? > I not the expert for coarse-graining of peptides, maybe somebody else > can comment on that, but if you average over both ensembles you will > get a force filed which possibly works in both cases. > > See: > Hills RD Jr, Lu L, Voth GA (2010) Multiscale Coarse-Graining of the > Protein Energy Landscape. PLoS Comput Biol 6(6): e1000827. > doi:10.1371/journal.pcbi.1000827 > and > Olga Bezkorovaynaya, Alexander Lukyanov, Kurt Kremer, Christine Peter > DOI: 10.1002/jcc.22915 > > > > > > It would be nice if you could post your script so that I can sort of > > create a list automatically and then work manually on the ranges. > I put it into one file: > #!/bin/bash > > beads=( A B C ) > echo "<cg>" > for ((i=0;i<${#beads[@]};i++)); do > for ((j=$i;j<${#beads[@]};j++)); do > cat <<EOF > <non-bonded> > <name>${beads[$i]}-${beads[$j]}</name> > <type1>${beads[$i]}</type1> > <type2>${beads[$j]}</type2> > <min>0</min> > <max>0.9</max> > <step>0.01</step> > <inverse> > <target>${beads[$i]}-${beads[$j]}.dist.tgt</target> > <gromacs> > <table>table_${beads[$i]}_${beads[$j]}.xvg</table> > </gromacs> > </inverse> > </non-bonded> > EOF > done > done > echo "</cg>" > > > > Thank you very much for your time. > > > > Cheers, > > Davide > > > > > > > > > > > > On 25/03/13 4:38 PM, "Christoph Junghans" <[email protected]> wrote: > > > >>2013/3/25 Mercadante, Davide <[email protected]> > >>> > >>> Dear Christoph, > >>> > >>> I am writing because I have a question about preparing my system for CG > >>>simulations using votca. > >>Welcome, and please if your research is not top-secret, post your > >>questions on our mailing list: <[email protected]>, so that > >>others can learn from it. > >>> > >>> I want to simulate a peptide in water using a CG approach and I was > >>>thinking of using votca to determine the non-bonded and bonded CG > >>>potentials starting from the full atom simulations that I have already > >>>performed on the same system (using Gromacs ersion 4.5.5). In order to > >>>do this I have studied the votca manual and I found it very interesting. > >>>I am interested to apply Boltzmann inversion method to find out bonded > >>>potential and IBI to determine the bonded potential in order to build a > >>>gromacs topology to use in further simulations. > >>> > >>> However, I have a question about calculating the non-bonded potential. > >>>It seems to me that I have to "feed" votca with a settings.xml file that > >>>lists all the non-bonded interactions occurring and other parameters to > >>>carry out IBI. However, in my system (not very small) the list of > >>>non-bonded interactions will be possibly very long..there is an > >>>automated way to build the settings file? How I can get a list of > >>>non-bonded interactions in my system? > >> > >>I am sorry to tell you that there is no automated way to create these > >>xml files, we thought about that several times, but as it can be done > >>with a 10 line shell script and most interactions will have different > >>ranges, which will need some tweaking anyhow, we never wrote a program > >>to do it. > >> > >>I can post my script if you need it! > >> > >>Cheers, > >> > >>Christoph > >> > >>> > >>> Thank you in advance for your time and please accept my apologies if > >>>this is a stupid question. I am a newbie of votca. > >>> > >>> Cheers, > >>> Davide > >>> > >>> ********************************************** > >>> Davide Mercadante Ph.D. > >>> Postdoctoral Fellow > >>> Heidelberg Institute for Theoretical Studies HITS > >>> Schloss-Wolfsbrunnenweg, 35 > >>> D-69118 Heidelberg, Germany. > >>> ********************************************** > >>> > >>> -- > >>> Christoph Junghans > >>> Web: http://www.compphys.de > > > > > > -- > 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 http://groups.google.com/group/votca?hl=en. > For more options, visit https://groups.google.com/groups/opt_out. > > > -- 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 http://groups.google.com/group/votca?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
