pseudochiral N or why not to develop too good Clean3D tools!

Hi - recently wrote a molecular overlay program seeking for optimal superpositions of a candidate compound atop of a reference. I'm using for input 3D sdfiles generated with Clean3D according to Ödon's Conformers.java (posted on this forum) - e.g. I get an .sd file where the first n1 "molecules" are conformations of my first compiound (the reference), the next n2 are conformers of molecule#2, and so on... I therefore need to know, when reading my way through the multiple conformer file, where a molecule stops (e.g. all its conformers were read) and a new one begins. I therefore systematically compare the smiles (obtained from mol.toFormat("smiles")) of the current input Molecule object to the one of the previously processed Molecule. If they differ, means that the conformer set of a new compound has begun at the current Molecule, otherwise I just picked another geometry of the current compound.





Unfortunately, R1-N-R2 amines, when adding hydrogens, become "chiral" (at 0 K) - it's either R1-N@H-R2 or R1-N@@H-R2 and, since this is serious soft and does things well, Clean3D would (and certainly does) enumerate BOTH!! Which sometimes makes my code believe that a conformer set's DONE, if suddenly it reads R1-N@@H-R2 instead of the expected R1-N@H-R2!





Now, I might use toFormat("smiles:0") to ignore stereo - but then, if I have two TRUE enantiomers following each other, my code'll think that they are mere conformers of a same compound! "smiles-H" won't do anything.





Big box of Martzipan for the discoverer of a trick for quick recognition of two conformers of a same species (well, yes, damn the atropoisomers)!

Hi!





We use the following rule to decide wheter a Nitrogen can be Chiral or not:


N is (under normal circumstances) not a stereocenter since it is flexible enough to invert, but if the atom itself is in ring and all its' ligands are also in ring, this flexibility vanishes.


As far as I understand you found a molecule for which this algorithm returns that the Nitrogen atom can be chiral, however it is not true.


Could you attach the molecule!





Thank you


Andras

You're right, guess my previous assesment of the problem was not overly accurate, for I was too lazy to VISUALIZE those smiles... turns out that the problem's a bit different from what I was assuming it to be. One of the faulty smiles is O=C1C2Cc3ccccc3CN2C(=O)N1CCCc4ccccc4, which, after conformational sampling, output to .sdf and molconvert smiles:-H yields TWO derived smiles





O=C1[C@@H]2Cc3ccccc3C[N@@H]2C(=O)N1CCCc4ccccc4





and





O=C1[C@@H]2Cc3ccccc3C[N@H]2C(=O)N1CCCc4ccccc4





You're right - the N in question IS in a ring... but it's a bloody AMIDE and I don't know what (the devil) that @(@)H sitting atop of it is doing there! - though the original .sdf (with all explicit H added!) produced by conformers.java does NOT display any H on those N, fortunately!





I also noticed, now that I've done my bug investigator job properly, that compounds with chiral carbons in the input geometry, for which stereochemistry is NOT SPECIFIED, Clean3D will generate BOTH enantiomers... which is actually no bug, but the right thing to do. It would, though, be cooler if the output conformers of the R and S species were grouped together (what if you'd generate N conformers for an arbitrary chirality and then add the N mirror images to the pool, in keeping the order 1...N,1'...N' - is this difficult to implement?)





Cheers - and Boldog Uj Evet if we don't talk again until then!

Hi!





The N should not have implicit H in this molecule.


So I can imagine the bug either in the conformation sampling or in the Smiles export.


Could you attach the generated sdfile for this molecule so I can check the export.





Odon will answer your question about the conformer generation.





Happy New Year!





Andras

dragos wrote:

I also noticed, now that I've done my bug investigator job properly, that compounds with chiral carbons in the input geometry, for which stereochemistry is NOT SPECIFIED, Clean3D will generate BOTH enantiomers... which is actually no bug, but the right thing to do. It would, though, be cooler if the output conformers of the R and S species were grouped together (what if you'd generate N conformers for an arbitrary chirality and then add the N mirror images to the pool, in keeping the order 1...N,1'...N' - is this difficult to implement?) Cheers - and Boldog Uj Evet if we don't talk again until then!

Hi Dragos,





Happy New Year (more details will follow in a separate message)!





Grouping the output of the conformer search for each epimer is doable, however, I do not really want to change the current way which simply based on energy for ordering, because it might lead to confusion. I have a few tips how to solve this externally and will explain it later.





Ithink that simply doing the smiles export with no explicite Hydrogens should do the trick, since the Nitrogene will loose chirality but the rest should keep it.





Best wishes and good luck,





Ödön