MoSS Atom Types

Hi All,

I'm processing a data set using MoSS. However it failed with a JavaNullpointer error because of 1 entry containing a deuterium isotope. (we try to avoid them but sometimes they slip through.)

Is this simply not supported or is it a bug?

Thanks,

Peter

peterem wrote:

I'm processing a data set using MoSS. However it failed with a JavaNullpointer error because of 1 entry containing a deuterium isotope. (we try to avoid them but sometimes they slip through.)

Is this simply not supported or is it a bug?


Smiles (and also SLN and SDF I guess) does not support deuterium "atoms", so MoSS simply does not support that.

Regards,

Thorsten

thor wrote:

Smiles (and also SLN and SDF I guess) does not support deuterium "atoms", so MoSS simply does not support that.

Regards,

Thorsten

Hi Thorsten,

Thanks for the quick reply.
SD file does support D or at least MDL does because it has not problems importing or exporting SD filescontaining this atom types into the MDL database. (so the databases also support it).

Anyway, Is there an way to filter them within Knime?

Thanks,

Peter

peterem wrote:
Thanks for the quick reply.
SD file does support D or at least MDL does because it has not problems importing or exporting SD filescontaining this atom types into the MDL database. (so the databases also support it).

Ah, good to know. I have to look into the Molfile spec some time.

peterem wrote:
Anyway, Is there an way to filter them within Knime?

Not that I can think of. However, I may ask the MoSS author to add support deuterium (and maybe also tritium). But from how I know the algorithm that can be a bit tricky. Usually hydrogens are ignored during frequent fragment mining, as they are not that important and make the search considerably more complicated. I'm not familiar with deuterium atoms, but I would assume they should be treated like hydrogens?

Regards,

Thorsten

peterem wrote:
Anyway, Is there an way to filter them within Knime?

Not that I can think of. However, I may ask the MoSS author to add support deuterium (and maybe also tritium). But from how I know the algorithm that can be a bit tricky. Usually hydrogens are ignored during frequent fragment mining, as they are not that important and make the search considerably more complicated. I'm not familiar with deuterium atoms, but I would assume they should be treated like hydrogens?

Regards,

Thorsten

Hi Thorsten,

Sorry, forgot about this forum item.
Anyway, I totally agree on you assumption that the deuterium atoms should be treated like hydrogens. Not sure if you need to change the code specially for this. Maybe a skip option for not handled structures is nice. I'm processing large sd-files and it takes quiet some time to figure out which entry caused the problem and also to rewrite the file again.

Cheers,

Peter