Re: [External] Boost value in aMD simulation

From: James Starlight (
Date: Fri Aug 22 2014 - 11:54:05 CDT

Thanks alot!

BTW I've tried to compare eigenvectors from each trajectory: for simlicity
I made 5 aMD identical 20-50ns calculations with double boost (each step ~
80 Kcal/mol have been added as the boost to dihedral) for my system
(restrained all atoms but not refined loops). In each case I only change
initial velocitiess for the system during the heating and varry
prolongation of the simulation (from 20 to 50 ns). Than I've made PCA of
each trajectory and compare eigenvectors (by the dot product ) with the
eigenvectors from the first (called it reference) longest trajectory. As
the result I 've obtained ~ 0.8 overlap in top 5 modes (with biggest
varience) . The only difference was in the shortest trajectory where only
5th (not 1 as in the rest cases) eigenvector had bigger overlap with 1st
eigenvector from reference trajectory ( in the rest cases the identical
pairs of the eigenvectors showed good correlation- in table it seems as the
linnear pattern of correlaiton).
 Does this indicate of good sampling and convergence in cases of all
trajectories (with the exeption of 2nd shortest trajectory)? How it could
be checked besides in terms of eigenvectors analysis? I really would like
to compare this with the cluster populations.

Thanks for suggestions,


2014-08-21 15:05 GMT+04:00 Thomas Evangelidis <>:

> On 19 August 2014 21:45, James Starlight <> wrote:
>> Thanks for suggestions!
>> Regarding simulation with the ligand: does the idea to perform clustering
>> of the receptor's regions (loop) being interacting with different ligands
>> seems good (e.g to compare results of apo-holo simulation to detect some
>> shared clusters between bpoth systems)?
> There are many things you can do. None can tell if something 's worth
> doing or not but you. Notwithstanding, I wouldn't look for common clusters
> but I would compare the predominant cluster representatives in the apo and
> holo simulation to identify induced fit conformational changes in the
> ligand binding region.
>> Regarding cut-off: when I reduced it to 1-2 A I still obtain 6-7 clusters
>> but most of the conformers were as the outliers (not in any of these
>> clusters).
> Vague statement. Cannot make conclusions without cluster numbers and
> populations.
>> Regarding dendrograms: for me better to find some python package for such
>> task :-) I'll try to check for it! Might the g_cluster be also usefull
>> within my taks besides of obtaining representative structures from each
>> clusters?
> This might help:

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:22:45 CST