Protonation corrupts some multiple-ring structures

Hello ChemAxon,

I am contacting you about a problem that we, at the bioinformatic research group at SRI International, find to be significant. I have spoken with people at ChEBI (Marcus Ennis) and UniProt (Anne Morgat) and they encounter the same hurdles.

When describing the stereochemistry of structures such as sterols, it is customary to draw several hydrogen bonds with stereo information (for example, look at the structure of cholesterol at wikipedia, which has 4 such hydrogens -see http://en.wikipedia.org/wiki/File:Cholesterol.svg

When the Marvin protonation plugin protonates such structures, it removes all of these hydrogens, and represents the stereo information that was defined by them by adding up and down directions to some of the wedges within the ring itself (see below). The structures end up looking so bad that we have stopped using Marvin for protonating any of the compounds that could be effected, and now perform the protonation of these compounds by hand. Similarly, curators at ChEBI manually fix every such compound following the analysis by Marvin.

Needless to say, these hydrogens all have pKa values that fall far beyond the relevant range, and really should not even be considered by the pKa module. Would it be possible to modify the protonation code so that it ignores hydrogens whose pKa falls outside the range of 1-14, and leaves those bonds as-is? This would be a great help to several of your "power users".

 

Thank you,

Ron Caspi

rcaspi wrote:

When the Marvin protonation plugin protonates such structures, it removes all of these hydrogens, and represents the stereo information that was defined by them by adding up and down directions to some of the wedges within the ring itself (see below). The structures end up looking so bad that we have stopped using Marvin for protonating any of the compounds that could be effected, and now perform the protonation of these compounds by hand. Similarly, curators at ChEBI manually fix every such compound following the analysis by Marvin.

Thanks for the feedback, we will consider implementing it.

rcaspi wrote:

Needless to say, these hydrogens all have pKa values that fall far beyond the relevant range, and really should not even be considered by the pKa module. Would it be possible to modify the protonation code so that it ignores hydrogens whose pKa falls outside the range of 1-14, and leaves those bonds as-is? This would be a great help to several of your "power users".

pKa plugin already has options to set the pKa range to consider. The options are:

Min basic pKa: widens the calculation range because weak bases will have lower pKa values than the default -10.

Max acidic pKa: widens the calculation range because weak acids will have higher pKa values than the default 20.

For more see http://www.chemaxon.com/marvin/help/calculations/protonation.html#pka

Zsolt

Dear Ron,

 

Thanks for reporting this problem! 

We will fix it asap.

 

Jozsi

Dear Jozsi,

Great. We are looking forward!

Take care,

Ron

Hi Jozsi,

looks like a fix didn't make it into beta 5.10. Do you have an idea when we could expect a fix for this problem?

Thanks,

Ron

Hi,

The fix will be available in 5.10.1.  It will be released in July.

Jozsi

Jozsi,

Looks like this problem is still around in 6.0.

Ron

Hi,

 

I have tried the 6.0 version on the Chemaxon home site. See the attached figure. It seems to work now.

A  new option was introduced "Keep explicit hydrogens".  But unfortunately we managed to hide it from users.

Please switch on the "Keep explicit hydrogens" option. Because  according to the default setting it is switched off (we will change the default).

Jozsi

Right, it works! I did not notice that option. Thanks!