Hi Nathan,

thank you very much for your help. I really appreciate all your replies.
I hope larger-scale calculations will be available for APBS in a soon future.

However, I am curious to know how much RAM APBS requires in respect to the grid dimensions specified in the input file.
For example, for a grid with dimensions 225 x 225 x 225 how much memory is required by the software?

Understanding this would be really useful to understand which limits for the calculations I can reach with, for example, 2GB, 4GB or 8GB of RAM.

Can you please address me to understand how APBS requires memory in respect to the grid dimensions and other factors included in the calculations?
That would be incredibly helpful.

Regards,
Davide


2011/2/9 Baker, Nathan <Nathan.Baker@pnl.gov>
Hi Davide --

I would recommend looking for institutional or national resources that you can access for your calculations.  I'm sorry that we don't currently have the capacity for those calculations on our servers.  We are working with some of the US supercomputer centers to see if we can provide web services for larger-scale APBS calculations.  I will keep the list updated on how this progresses.

Sincerely,

__________________________________________________
Nathan Baker, Ph.D.
Pacific Northwest National Laboratory
Tel:  509-375-3997

From: Davide Mercadante <dmer018@aucklanduni.ac.nz<mailto:dmer018@aucklanduni.ac.nz>>
Date: Tue, 8 Feb 2011 16:15:28 -0800
To: Nathan Baker <nathan.baker@pnl.gov<mailto:nathan.baker@pnl.gov>>
Cc: APBS List <apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net>>
Subject: Re: [Apbs-users] Calculated binding energy seems to be too high?!

Hi Nathan,

thanks for your mail.
So, seems that even using the client I won't be able to increase the resolution of the grid. Do you have any other suggestions to solve the problem and increase the accuracy for the binding energy calculations?

There is any other resources/methods to perform such calculations?
This seems mainly to be a problem of size. Is this system too bid to be calculated?

Thank you.

Davide

2011/2/8 Baker, Nathan <Nathan.Baker@pnl.gov<mailto:Nathan.Baker@pnl.gov>>
Hi Davide --

You are using the client correctly; our server node may not have enough memory either.

APBS memory usage scales as approximately 120 B per grid point with a small amount of additional overhead for the molecule setup.

Thanks,

Nathan

________________________________
From: Davide Mercadante <dmer018@aucklanduni.ac.nz<mailto:dmer018@aucklanduni.ac.nz>>
To: Baker, Nathan
Sent: Sun Feb 06 17:43:45 2011

Subject: Re: [Apbs-users] Calculated binding energy seems to be too high?!

Hi Nathan,

thank you for your reply.

Yes, I have tried with the APBS opal client by using the Apbsclient.py command as specified in the documentation.

I have done several trials but even if I increase my dime to 257 x 257 x 257 (the next nlev) it gives me a memory allocation problem.
Howe much RAM I have to run calculations via OPAL?

Hope I am not doing something wrong...the command I use is:

ApbsClient.py --non-blocking {input file}

Please, can you address me to solve this problem? How much RAM I would need to increase the accuracy of my calculations?

Thank you infinitely.

Regards,
Davide

2011/2/7 Baker, Nathan <Nathan.Baker@pnl.gov<mailto:Nathan.Baker@pnl.gov>>
Hi Davide --

Have you tried submitting it using the APBS Opal client?

Thanks,

__________________________________________________
Nathan Baker, Ph.D.
Pacific Northwest National Laboratory
Tel:  509-375-3997

Date: Sat, 5 Feb 2011 13:39:15 -0800
To: Nathan Baker <nathan.baker@pnl.gov<mailto:nathan.baker@pnl.gov><mailto:nathan.baker@pnl.gov<mailto:nathan.baker@pnl.gov>>>
Cc: APBS List <apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net>>>
Subject: Re: [Apbs-users] Calculated binding energy seems to be too high?!

Dear Nathan,

thanks for your mail. Unortunately I don't have any possibilities to access a machine that has more RAM.

There is any server services for APBS where I can submit my input and pqr files to run the calculations using more memory?

Any helps would be greatly appreciated.

Thank you in advance.

Cheers,
Davide

2011/2/5 Baker, Nathan <Nathan.Baker@pnl.gov<mailto:Nathan.Baker@pnl.gov><mailto:Nathan.Baker@pnl.gov<mailto:Nathan.Baker@pnl.gov>>>
Hi Davide --

It appears that you're performing the calculation correctly; perhaps the grid isn't fine enough?  Do you have access to a machine where you can run a higher resolution (more memory) calculation?

__________________________________________________
Nathan Baker, Ph.D.
Pacific Northwest National Laboratory
Tel:  509-375-3997

Date: Thu, 3 Feb 2011 17:06:52 -0800
To: Nathan Baker <nathan.baker@pnl.gov<mailto:nathan.baker@pnl.gov><mailto:nathan.baker@pnl.gov<mailto:nathan.baker@pnl.gov>><mailto:nathan.baker@pnl.gov<mailto:nathan.baker@pnl.gov><mailto:nathan.baker@pnl.gov<mailto:nathan.baker@pnl.gov>>>>
Cc: APBS List <apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net>><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net>>>>
Subject: Re: [Apbs-users] Calculated binding energy seems to be too high?!

Dear Nathan,

thanks for your quick reply.
My pqr is simply the concatenation of the two monomers so that the monomers separately can be overlapped to the monomers in the dimer.pqr file.

However, I must say that I am centering the grid for each calculation on the dimer.
Should I center the grid for the calculations of each monomers on the actual monomers?

Thank you for your help.

Regards,
Davide

2011/2/4 Baker, Nathan <Nathan.Baker@pnl.gov<mailto:Nathan.Baker@pnl.gov><mailto:Nathan.Baker@pnl.gov<mailto:Nathan.Baker@pnl.gov>><mailto:Nathan.Baker@pnl.gov<mailto:Nathan.Baker@pnl.gov><mailto:Nathan.Baker@pnl.gov<mailto:Nathan.Baker@pnl.gov>>>>
Hi Davide --

This seems likely to be a problem with grid convergence.  Is your dimer.pqr file simply a concatenation of the two subunits or do the subunits change configuration/location/orientation on the grid?  If it is a simple concatenation, then your direct calculation represented by the first "print energy" line in your input file should work.  If not, you will have to construct a more complicated free energy cycle, adding properly-scaled coulombic interactions to the solvation energies in your calculation.

Good luck,

__________________________________________________
Nathan Baker, Ph.D.
Pacific Northwest National Laboratory
Tel:  509-375-3997

From: Davide Mercadante <dmer018@aucklanduni.ac.nz<mailto:dmer018@aucklanduni.ac.nz><mailto:dmer018@aucklanduni.ac.nz<mailto:dmer018@aucklanduni.ac.nz>><mailto:dmer018@aucklanduni.ac.nz<mailto:dmer018@aucklanduni.ac.nz><mailto:dmer018@aucklanduni.ac.nz<mailto:dmer018@aucklanduni.ac.nz>>><mailto:dmer018@aucklanduni.ac.nz<mailto:dmer018@aucklanduni.ac.nz><mailto:dmer018@aucklanduni.ac.nz<mailto:dmer018@aucklanduni.ac.nz>><mailto:dmer018@aucklanduni.ac.nz<mailto:dmer018@aucklanduni.ac.nz><mailto:dmer018@aucklanduni.ac.nz<mailto:dmer018@aucklanduni.ac.nz>>>>>
Date: Wed, 2 Feb 2011 02:05:20 -0800
To: APBS List <apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net>><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net>>><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net>><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net><mailto:apbs-users@lists.sourceforge.net<mailto:apbs-users@lists.sourceforge.net>>>>>
Subject: [Apbs-users] Calculated binding energy seems to be too high?!

Dear APBS community,

I have been using APBS to calculate the binding energy of a protein dimer by using sequential focusing solving PBE.
I have prepared an input file which compute the electrostatic energy for the dimer solvated (pdie 4 and sdie 78.5) and in vacuum (pdie and sdie = 4.0) and the two monomers composing the dimer in the same conditions.

I have performed a first coarse calculation (bcfl = zero) with a large map of lenght 300 x 300 x 300 A^3 and two focusing calculations with a medium map (length 100 x 100 x 100 A^3) and small map (length 75 x 75 x 75 A^3). I am using a "dime" of 225 x 225 x 225 that is the maximum allowed by the amount of RAM I have for the calculations...
I am attaching the input file and the pqr files in this mail.

However, the output file (attached as out.out) gives me a strangely high value for the global electrostatic energy (1.009E+04 Kj/mol.)
Is this correct? It seems to be much higher in respect of other values for complexes binding energies.
To calculate the binding I have used the energy cycle described in the APBS documentation. I have subtracted the solvated energies calculated for the monomer A and B to the actual energy of the solvated dimer...

Can you please advice me if there is something wrong or something that I could improve in my calculations? Or you think that the value I am getting for the electrostatic energy is unreasonably too high?

All the files are attached and please do not hesitate to ask if you need any other information.

Any help will be incredibly appreciated.

Thank you infinitely in advance.

Regards,
Davide

--
Davide Mercadante - PhD student -
Department of Chemistry
The University of Auckland
Rm 438 ext. 89171
1142 Auckland, New Zealand



--
Davide Mercadante - PhD student -
Department of Chemistry
The University of Auckland
Rm 438 ext. 89171
1142 Auckland, New Zealand



--
Davide Mercadante - PhD student -
Department of Chemistry
The University of Auckland
Rm 438 ext. 89171
1142 Auckland, New Zealand



--
Davide Mercadante - PhD student -
Department of Chemistry
The University of Auckland
Rm 438 ext. 89171
1142 Auckland, New Zealand



--
Davide Mercadante - PhD student -
Department of Chemistry
The University of Auckland
Rm 438 ext. 89171
1142 Auckland, New Zealand



--
Davide Mercadante - PhD student -
Department of Chemistry
The University of Auckland
Rm 438 ext. 89171
1142 Auckland, New Zealand