Problems with Waterswap 2016.1.0

[Emilio Lence Quintana]

Dear Sirs.
I downloaded and install the executable sire_16_1_linux64.run. I used gcc/6.1.0 and cmake/3.6.1, everything seemed to be fine. I run the test after installation and ended with a "ALL TESTS PASSED :-)", but also said that one or more jobes in OpenMM failed. I did not expect this to be a mayor issue, so I started some Waterswap calculations. They run fine, no problems apparently, until I checked the generated pdbs in order to see if the substitution process was OK, but then I discover that my ligand and the amino acids are distorted. Trying to find the origin of the problem, I found that the initial co-ordinates are fine, but the initially generated "bound_fixed.pdb" file, has similar distortions on the amino acids, for example, aromatic rings instead of flat have a chair conformation.
The files used, came from a MD simulation with AMBER14, ff14SB and gaff were used as forcefield for protein and ligand respectively. To overcome the waterswap limitation to truncated octahedron water boxes, I converted the different set of rst files to a cubic box. I do not expect this to be a problem, as the waters added are far from the ligand and form part of the fixed part of the system during Waterswap calculations.
My question is why this is happening? Do I need some more things to load before installation in order to pass the OpenMM test?
Probably I should try to download and install the source code, and hope for this solving the problem.
Thanks for your kind attention.
Best wishes
Emilio.

[Christopher Woods]

  Hi Emilio,

That's a strange problem. You don't need OpenMM to run waterswap or ligandswap, so do not work about OpenMM not working (there is an ongoing problem with OpenMM with newer GCCs that I have to solve).

waterswap/ligandswap work by cutting out a sphere of molecules/residues around the ligand to be swapped. Only molecules/residues within this sphere are moved. The remaining atoms outside the sphere are kept fixed, but they are still seen by the moving atoms. The fixed atoms from the protein box are all placed in the bound_fixed.pdb file. They are not moved at all by waterswap, so I am surprised that the residues showed distortion in the rings. This implies that the distortion was already there in the rst/top files you loaded into waterswap. Would you mind sending me the top/rst files you are loading, as this sounds like a very strange bug.

  Thanks,

  Christopher

Emilio.

--
You received this message because you are subscribed to the Google Groups "Sire Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sire-users+unsubscribe@googlegroups.com.
To post to this group, send email to sire-...@googlegroups.com.
Visit this group at https://groups.google.com/group/sire-users.
For more options, visit https://groups.google.com/d/optout.

--

---------------------------------------------------------
Christopher Woods
+44 (0) 7786 264562
http://chryswoods.com

[Emilio Lence Quintana]

Hi Christopher
I checked all the top/rst files an no distortion is observed, neither with pymol or vmd, on windows or linux. All look fine. The distortion is in all the residues except water, I mentioned the case of 6-membered aromatic rings because is the most easy to explain, but it happens in the ligand and all the amino acids. I sent attached the topology and coordinate files, and a couple of pdb so you can see the distortion. I don not thing geometry is the problem because I tried a short run on a installation of sire_14_4 (the one I used in Bristol when I was there) and no distortion is observed in the "bound_fixed.pdb" file.
I will try to install using the source code and see what happens. I think I did something wrong with my installation.
Thanks.
Emilio

On Tuesday, 20 September 2016 14:31:59 UTC+2, Christopher wrote:

  Hi Emilio,

That's a strange problem. You don't need OpenMM to run waterswap or ligandswap, so do not work about OpenMM not working (there is an ongoing problem with OpenMM with newer GCCs that I have to solve).

waterswap/ligandswap work by cutting out a sphere of molecules/residues around the ligand to be swapped. Only molecules/residues within this sphere are moved. The remaining atoms outside the sphere are kept fixed, but they are still seen by the moving atoms. The fixed atoms from the protein box are all placed in the bound_fixed.pdb file. They are not moved at all by waterswap, so I am surprised that the residues showed distortion in the rings. This implies that the distortion was already there in the rst/top files you loaded into waterswap. Would you mind sending me the top/rst files you are loading, as this sounds like a very strange bug.

  Thanks,

  Christopher

On 20 September 2016 at 12:38, Emilio Lence Quintana <emilio.len...@gmail.com> wrote:

Dear Sirs.
I downloaded and install the executable sire_16_1_linux64.run. I used gcc/6.1.0 and cmake/3.6.1, everything seemed to be fine. I run the test after installation and ended with a "ALL TESTS PASSED :-)", but also said that one or more jobes in OpenMM failed. I did not expect this to be a mayor issue, so I started some Waterswap calculations. They run fine, no problems apparently, until I checked the generated pdbs in order to see if the substitution process was OK, but then I discover that my ligand and the amino acids are distorted. Trying to find the origin of the problem, I found that the initial co-ordinates are fine, but the initially generated "bound_fixed.pdb" file, has similar distortions on the amino acids, for example, aromatic rings instead of flat have a chair conformation.
The files used, came from a MD simulation with AMBER14, ff14SB and gaff were used as forcefield for protein and ligand respectively. To overcome the waterswap limitation to truncated octahedron water boxes, I converted the different set of rst files to a cubic box. I do not expect this to be a problem, as the waters added are far from the ligand and form part of the fixed part of the system during Waterswap calculations.
My question is why this is happening? Do I need some more things to load before installation in order to pass the OpenMM test?
Probably I should try to download and install the source code, and hope for this solving the problem.
Thanks for your kind attention.
Best wishes

Emilio.

--
You received this message because you are subscribed to the Google Groups "Sire Users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to sire-users+...@googlegroups.com.

To post to this group, send email to sire-...@googlegroups.com.
Visit this group at https://groups.google.com/group/sire-users.
For more options, visit https://groups.google.com/d/optout.

[Emilio Lence Quintana]

Dear sire users.

The plot thickens.

I left late yesterday so I couldn't reinstall sire from source code, but today in the morning I found that some waterswap jobs that I had on queue started, and surprisingly when I checked then, the pdb files generated were OK. So, it seems there is not a problem with the installation. And the initial top and rst files are Ok and they are generated the same. I'm still waiting for one of the failed jobs to start from the beginning and see if the calculation goes wrong again.

I thought that maybe there was some problem between the gcc/6.1.0 that I need to load in the machine to run waterswap, with the modules intel/2016  impi/5.1 cuda/7.5 amber/14-at15 that I loaded before in order to generate the topology and coordinate files with a cubic water. I contacted IT form the supercomputing I'm using, and they said that this should not be the problem, because when gcc is loaded the intel module is unloaded.

I check the output file generated from the calculation to see if it sheds any lignt in what is hapenning here. The first thing that I found is a lot of lines like these:

HOW IS THE MOLECULE OUTSIDE THE SPHERE?
1.42611 1

But I think that this is not a problem, and I heve some cases with lines like that before witho any problem.

Also I found this in all calculations:

Test replicas 0 1
beta 1.6878 1.6878
i -6490.51 -6479.45 11.0597
j -6522.97 -6527.56 -4.58187
Passed? false
Test replicas 2 3
beta 1.6878 1.6878
i -6499.13 -6491.17 7.96458
j -6510.76 -6518.66 -7.89575
Passed? true
Test replicas 4 5
beta 1.6878 1.6878
i -6463.2 -6452.18 11.02
j -6470.85 -6478.38 -7.53775
Passed? false
Test replicas 6 7
beta 1.6878 1.6878
i -6462.21 -6447.91 14.304
j -6453.56 -6458.04 -4.48021
Passed? false
Test replicas 8 9
beta 1.6878 1.6878
i -6441 -6429.38 11.621
j -6431.37 -6438.88 -7.51368
Passed? false
Test replicas 10 11
beta 1.6878 1.6878
i -6426.08 -6413.95 12.1282
j -6455.62 -6464.28 -8.66261
Passed? false
Test replicas 12 13
beta 1.6878 1.6878
i -6412.36 -6402.72 9.64181
j -6443.24 -6451.3 -8.05925
Passed? false
Test replicas 14 15
beta 1.6878 1.6878
i -6376.89 -6363.96 12.9305
j -6391.11 -6398.61 -7.5055
Passed? false

These lines sometimes appear right at the beginning of the file, and always a lot of times in the middle of the "HOW IS..." lines, counting the word Test, it appears more than 7000 times. I check calculations that I did before in the 14_4 verison of sire and there is no trace of that lines in them. So either is part of the problem, or is a new feature of the 2016 version.

Apart from that, the rest of the file looks normal.

I will keep you informed of any development.

Best wishes

Emilio Lence.

 

--
You received this message because you are subscribed to a topic in the Google Groups "Sire Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/sire-users/cpOsZx6AsaI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to sire-users+unsubscribe@googlegroups.com.

To post to this group, send email to sire-...@googlegroups.com.
Visit this group at https://groups.google.com/group/sire-users.
For more options, visit https://groups.google.com/d/optout.

--

Emilio José Lence Quintana
Post-doctoral researcher
Phone +34 8818 15759
Centro Singular de Investigación en Química Biolóxica y Materiais Moleculares (CiQUS)
C/ Jenaro de la Fuente s/n (esquina Avda. Mestre Mateo)
Campus Vida
Universidade de Santiago de Compostela
15782 Santiago de Compostela
Spain

[Christopher Woods]

  Hi Emilio,

There messages about the sphere and the replica exchange tests can be ignored, as they relate to debugging information that I left in the production code. I agree with your IT person that the other modules that were loaded should not have affected the waterswap calculation, or caused the issue you saw. 

This is an extremely strange bug. There is no code in Sire that would even know how to change a flat ring into a chair conformation, and there is definitely no code that moves the atoms in the bound_fixed.pdb (they should be identical to what was read from the rst/top files). 

If this happens again, could you send me the wsrc_restart.s3 file so I can inspect the coordinates held within there? (or even, send me a tgz containing the entire directory with all results). This seems to be a freak occurrence, but I would be happier if I can track down exactly why this has happened.

  Cheers,

  Christopher

p.s. Unfortunately I am going away on holiday tomorrow and have no time today to look at this further. I am afraid that I won't be able to look into this more until I get back in a few weeks.

To unsubscribe from this group and stop receiving emails from it, send an email to sire-users+unsubscribe@googlegroups.com.

To post to this group, send email to sire-...@googlegroups.com.
Visit this group at https://groups.google.com/group/sire-users.
For more options, visit https://groups.google.com/d/optout.

--

[Emilio Lence Quintana]

Hi Chris.

Thanks for your answer. We are trying several things, installation from the source code, installation of older versions in another machines, etc., to try to fine were the problem is.

For sending the files, do you prefer that I put then in curie/bc3, becuase I still have access to those machines.

Best wishes, Emilio

P.S.: Enjoy your holidays.

On 22 September 2016 at 12:17, Christopher Woods <chrys...@gmail.com> wrote:

  Hi Emilio,

There messages about the sphere and the replica exchange tests can be ignored, as they relate to debugging information that I left in the production code. I agree with your IT person that the other modules that were loaded should not have affected the waterswap calculation, or caused the issue you saw. 

This is an extremely strange bug. There is no code in Sire that would even know how to change a flat ring into a chair conformation, and there is definitely no code that moves the atoms in the bound_fixed.pdb (they should be identical to what was read from the rst/top files). 

If this happens again, could you send me the wsrc_restart.s3 file so I can inspect the coordinates held within there? (or even, send me a tgz containing the entire directory with all results). This seems to be a freak occurrence, but I would be happier if I can track down exactly why this has happened.

  Cheers,

  Christopher

p.s. Unfortunately I am going away on holiday tomorrow and have no time today to look at this further. I am afraid that I won't be able to look into this more until I get back in a few weeks.

[Christopher Woods]

Yes, putting them on curie or bc3 would be great.

  Cheers,

  Christopher

--
Sent from Gmail Mobile