[gmx-users] pca-based MD

[James Starlight]

Thomas,

thank you for the explanation

1) Indeed ED sampling was exactly that I need. It's not quite
understand for me about correct chose of that parameters for biassing
simulation

-linfix      string         Indices of eigenvectors for fixed increment
                            linear sampling
-linacc      string         Indices of eigenvectors for acceptance linear
                            sampling
-radfix      string         Indices of eigenvectors for fixed increment
                            radius expansion
-radacc      string         Indices of eigenvectors for acceptance radius
                            expansion
-radcon      string         Indices of eigenvectors for acceptance radius
                            contraction

What exactly I need is the biasing of the new MD run along one
specified eigenvector which I've extracted from previous run. So it's
not quite understand for me what chose would be most correct e.g
linnear sampling along that eigenvector or rafius expansion.

2) Also I want to perform such biassing along PC mode extracted from
the X-ray data-set of my protein. E.g on first step I calculate PC
from the X-ray data-set consisted of different functional relevant
conformations of my protein ( for this purpose I need create
trajectory from that x-ray data-set firstly). Than I'm looking for one
PC which correspond to biological relevant motion ( e.g opening of
active center seen in first PC). Finally I want to perform MD from one
start structure ( e,g its closed form) along extracted 1 PC from x-ray
data-set. For that step I also must make *.edi file for further md_run
of the protein firslty. Does this workflow correct in general ? :)


3) This FMA technique is very intresting indeed. As I've understood It
can be usefull in case where I could not find biological-relevent
motion along one specified PCs ( In that case the combination of the
PCs might represent this motion which can correspond to one
eigenvector made by FMA). Does it correct ?

Thanks again for help,

James
2012/9/23 Thomas Evangelidis <tevang3 at gmail.com>:
> I presume you are referring to Essential Dynamics Sampling, described in
> section 3.14 of the manual (v4.5.4). There is also a great tool that finds
> the few PCs that are maximally correlated to a functional quantity (e.g.
> the volume of the active site). The technique is coined Functional Mode
> Analysis (FMA) and you can find more information at:
>
> http://xray.bmc.uu.se/~jochen/fma.html
>
> I have used FMA and worked pretty well in my case. I am wondering if anyone
> thought of using that technique to find the PCs that are maximally
> correlated to a functional quantity and then perform Essential Dynamics
> sampling  on these PCs to explore the conformational space that affects the
> most that functional quantity.
>
> I.e. I am studying a kinase in the wt and mutant form. Although the
> mutation is not near the active site there is a lot of discussion in the
> literature about the effect of the mutation on the opening of the catalytic
> cleft. Some people claim that one possible explanation of the over-activity
> of the mutant is the greater opening of the active site, which facilitates
> substrate binding and thus leads to enhanced reaction turn-over. In order
> to test this hypothesis with unbiased MD one would need tremendous computer
> resources and a lot of time (the kinase is gigantic). On the other hand one
> could run short simulations of the wt and mutant, do FMA to find the 10-20
> PCs that are maximally correlated to the volume of the active site, and
> then perform Essential Dynamics Sampling on these PCs to explore the
> conformational space that is highly correlated to the volume of the active
> site. After that, one could safely claim that the Hypothesis was true or
> false.
>
> I would be interested to read your comments on this.
>
> Thomas
>
>
> On 23 September 2012 11:19, James Starlight <jmsstarlight at gmail.com> wrote:
>
>> Dear Gromacs Users!
>>
>>
>> There are many publications about implementation of the pca-based MD
>> simulations for the investigation of the functional-relevant motions.
>> In that cases the eigenvectors are extracted from the relatively short
>> MD simulation of the investigated protein and than the biassed MD
>> simulation is started along chosen principal component which used as
>> the reaction coordinate.
>>
>> I'd like to know more about implementation of that technique in
>> Gromacs. E.g if I've performed some PCA and extracted eigenvectors how
>> I can run further simulation along one of the chosen PC ?
>>
>> Thanks for help
>> James
>> --
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>> * Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-users-request at gromacs.org.
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>
>
>
> --
>
> ======================================================================
>
> Thomas Evangelidis
>
> PhD student
> University of Athens
> Faculty of Pharmacy
> Department of Pharmaceutical Chemistry
> Panepistimioupoli-Zografou
> 157 71 Athens
> GREECE
>
> email: tevang at pharm.uoa.gr
>
>           tevang3 at gmail.com
>
>
> website: https://sites.google.com/site/thomasevangelidishomepage/
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists