Inconsistent pKa calculations on polyphosphate moieties

This post is a follow-up to the following thread:

http://www.chemaxon.com/forum/ftopic4220.html&view=next

Thank you for all of your previous help with using Marvin, and for providing the scientific community with such a wonderful tool.

I wanted to report today a problem that we have been experiencing with the Marvin pKa plug-in, again with phosphate moieties. We have noticed that compounds with polyphosophate moieties are generating different pKa values for the terminal phosphate depending on how the oxygen-phosphorus double-bonds are aligned between adjacent phosphates. This is troublesome for us, because these differences are straddling pH = 7.3, which is the reference pH which we are using for normalizing all of our chemical structures for protonation (i.e., the major microspecies).

For example, we have scrutinized the compound dihydromonampterin triphosphate. I have attached two images of the same molecule,w ith the only difference being the orientation of the double-bonds in the phosphates relative to one another. In other words, in one picture, the double-bonds of the phosphates are all on the "bottom side" of the structure image, whereas in the other picture the double-bonds are not aligned. The aligned version produces a pKa of 7.419 on the terminal phosphate, whereas the unaligned version of the same structure produces a pKa of 6.819 on the terminal phosphate.

Considering that the placement of single and double bonds in a phosphate are just approximations to the true resonance structure, my colleages and I do not believe that the pKa calculation for the terminal phosphate should be so sensitive to the placement of double-bonds in the polyphosphate tail. In order to make our database more consistent, we are asking for the Marvin team to try to determine the source of this error, and to let us know when a fixed version is available.

Thank you very much.

Kind regards,

~Tomer Altman

Bioinformatics Research Group

SRI, International

MetaCyc.org EcoCyc.org HumanCyc.org BioCyc.org

Hi,

 

I have tested your molecule with the Marvin 5.3 version and I was not able to reproduce this error.

Marvin 5.3 will be available at the begining of January.

 

Could you send the two molecules in any struture file format like "mol" or "sdf"?

 

Thanks!

 

Jozsi

Hi Joszi,

I tested this using MarvinSketch 5.2.06:

Product Version:
MarvinSketch 5.2.6
Build Date:
2009-10-19
Operating System:
amd64 Linux 2.6.15-54-amd64-server
Java:
Sun Microsystems Inc. Java 1.6.0
Memory:
448.0M maximum, 136.4M total, 37.8M free
Environment:
Application
Browser:

 

My understanding is that is the current production version of Marvin. Can you verify for me that this problem exists using MarvinSketch 5.2.06?

 

I've attached our MOLfile.

 

Thanks,

 

~Tomer

Hi,

 

I was able to reproduce the bug with the molecule you attached. I will fix this bug asap. The fixed version of Marvin 5.3 will be availble at the begining of January.

 

Thank you!

Jozsi

Hi Jozsi,

Thanks for verifying the bug. To clarify, did you reproduce the bug in the released version (5.2.06), or in the upcoming 5.3 version?

Thank you,

~Tomer

Hi,

 

I just checked in the 5.3 version.

 

Jozsi

Hi Jozsi,

Is the bug now fixed?

Thanks,

 

~Tomer

Hi,

 

Yes, it is fixed in 5.3.

 

Jozsi