mcs for a set of molecules

Hi,

does chemaxon has implemented some class to find (as much as possible) the (approximate) MCS for a set of molecules, not just two? I think libMCS does that "behind the scene" to generate a hierarchy, but it would be great to tap on that in order to find approximate MCSs for sets of molecules.

Thanks,

Gonzalo

Hi Gonzalo,

You can tweak LibraryMCS to generate one level of the hierarchy: setAllowedLevelCount(2).

Regards

Miklos

Can you get these from libmcs command? What do the two columns with integers represent when you save the output in CSV format?
Thanks

I'm afraid you can't control that in batchmode, however you can change in the the GUI.

Regarding the two values in the CSV output:

1, the so-called hierarchy ID, that is, the level number, from top to bottom, top level nodes are on hierarchy level 1

2, the second is the index of the node on its level from the left, left most node index  is 1

Well, I'm not sure these would be particularly useful indices in your application. We did this on specific user request. CSV is important but perhaps other internal references could better support the handling of clustering results in third party applications. We haven't revised this internally yet.

What's your opinion on that? What would be the meaningful CSV output for you?

Thanks

Miklos

Oh, now I understand. The current output then seems sensible to me; maybe an additional column with an integer indicating the number of compounds being represented by that MCS would be useful (as it appears in the GUI).

So from what you say, although you can't control the level of MCS in the output of command line libmcs, you actually obtain all of the levels in the hierarchy in the CSV output file, including both the intermediate MCSs (with increasing complexity as you move throgh higer levels in the hierarchy) and (at the highest level) the actual molecules. Is that correct?

 

Thanks,

Gonzalo

 

Oh, now I understand. The current output then seems sensible to me; maybe an additional column with an integer indicating the number of compounds being represented by that MCS would be useful (as it appears in the GUI).

This sounds reasonable, we do that.

So from what you say, although you can't control the level of MCS in the output of command line libmcs, you actually obtain all of the levels in the hierarchy in the CSV output file, including both the intermediate MCSs (with increasing complexity as you move throgh higer levels in the hierarchy) and (at the highest level) the actual molecules. Is that correct?

Yes, that's correct.

 

Miklos