Umbrella sampling simulations using the distance between two centers of masses as reaction coordinate

[Vasileios Tatsis]

Dear PLUMED-users,

I performed umbrella sampling (US) simulations, using Gromacs 4.6 and Plumed 2.0, of a biomolecule in explicit solvent.

All  US simulations start from the folded structure of the biomolecule. As a reaction coordinate I use the distance between two centers of masses.

The US simulations have a length of 100 ns each and the distance (CV1) between successive US windows is 0.1nm, spanning distances between 1.0 nm and 4.0 nm.

For the calculation of the PMF I use the WHAM code and I discard the first 20 ns. 

I used three different spring constants for the harmonic potential to restrain the distance CV1: 500 KJ/mol/nm^2, 1000 KJ/mol/nm^2,  2000 KJ/mol/nm^2.

Also, the parameters of the Plumed input file, making sure also to use the WHOLEMOLECULES keyword, are the following:

restraint-d1: RESTRAINT ARG=d1 KAPPA=500 AT=1.0

PRINT STRIDE=10 ARG=d1,restraint-d1.bias FILE=COLVAR-d1.0.dat

I plotted the histograms for each of the US windows for each simulation and they seem to have a proper overlap.

The PMF resulting from these three US simulations is totally different for each case.

I attach to this message the histogram plots and the PMF plot.

Thanks in advance for your comments.

Vassilis

[davide branduardi]

Hi, 

Just a word, if you want to measure the folding of a biomolecule, it is difficult that the end-to-end distance will give you the right result (especially if you test backwards, which is something that one should do).

Assuming that the CV choice is not a problem (but generally it is the problem)

for any of those, do you obtain the same profile if you use half of the statistics?

In other words, are your distributions converged? They look so, but you never know: you have some logs in the formula to extract the pmf that might give a non intuituve result. 

Additionally, do you see your simulations crashing? If so, then it might well be that the spring constant is too hard.

And lastly, as said, if you try for any of the spring constant, to do the process reversed and you get another different result, then your CV choice is bad.

Just some ideas from experience...

Good luck!!!

D

******************************************************
Davide Branduardi
davide.b...@gmail.com
Web:
https://sites.google.com/site/davidebranduardi
******************************************************

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.
To post to this group, send email to plumed...@googlegroups.com.
Visit this group at http://groups.google.com/group/plumed-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/4be9e4eb-dc07-4b66-8bcb-00777201f8f2%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Vasileios Tatsis]

Dear Davide,

thank you for the swift response,

i ) indeed, I tested the convergence of my results by extending the simulation time to 200ns for the case of 1000 KJ/molnm^-2, which is reached after 100ns. 

ii) None of the three sets of US simulations reported a crash.

iii) By reverse process, you mean starting from the unfolded structure and perform the same procedure?

Thanks again,

V.

[davide branduardi]

Yes,

I mean starting from the unfolded. Consider that this is rather standard assessment in the field.

And for convergence, I mean convergence in each single umbrella: each "bell shaped" distribution should be converged in time.

Nevertheless my gut feeling is that for folding the end-to-end distance is not  a good CV....

Davide 

[Vasileios Tatsis]

Dear Davide,

I started, als,o using the radius of gyration as a reaction coordinate for the (un)folding.

According to your experience, which other reaction coordinate(s)/collective variable(s) are suitable for the examination of (un)folding process?

Vassilis

[davide branduardi]

Hi !

People used a lot of different stuff.

One of the work that I like is PT-Metad work from bussi gervasio et al.

They have gyration radius and hbond which are rather general but deficient and they had to put on top metad and PT to overcome the other barriers that you might have from other effects (e.g. solvation?) but that you can overcome with temperature from the PT. 

Probably Giovanni can comment more on this.

Another nice development is the WTEnsemble from Max Bonomi. He made some cool stuff that you can dig out in the recent literature. 

My advice is to read the most recent literature on it, give a try and make your mind about it. There is no "right" cv. Every system is a different thing.

Good luck!

D

******************************************************
Davide Branduardi
davide.b...@gmail.com
Web:
https://sites.google.com/site/davidebranduardi
******************************************************

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.
To post to this group, send email to plumed...@googlegroups.com.
Visit this group at http://groups.google.com/group/plumed-users.

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/5a8530d3-9e59-4d6f-8414-c87de8b4330b%40googlegroups.com.