On Wed, Oct 13, 2010 at 8:58 PM, Geoffrey Hutchison <ge...@geoffhutchison.net> wrote: >> 3429680: [Li+251] 24639246 >> 3429701: [ClH+276] 24639289 >> 3429702: CCCC[n+251]1cccc(C)c1 24639291 > > These are an easy fix to the SMILES reader. Currently, we only take single > digits for charge. Now I think the molecule is preposterous (e.g., lithium > doesn't have 251 electrons to remove!) but I can certainly imagine some > metals with formal charges of +10 or above. > > My feeling is that eMolecules probably wants to add some level of filtering > to remove species with highly unusual valence.
Yes, but we should handle this by throwing an error. There are probably more of these cases though. > Also, does the SMILES format have a "continue after error" flag like SDF? > That seems like a good addition. I don't think so but this would be nice since it is a widely used format. It would also be nice to have an option to write all molecules that produced an error to a file for example. (Gert suggested this) This could be done from OBConversion I think. >> These have large rings which are not found I think. We should be able >> to correctly detect ring membership though since this is done using a >> spanning tree before SSSR/LSSR analysis is done. I'll take a look at >> this. > > I found this similar bug when doing fragment analysis last night. I have a > defensive fix which solves the segfault but does not fix the analysis. I'll > commit that shortly which is useful anyway. Sounds good. >> consider the H atoms that are added when writing out the smiles. I can >> add this to the canonical code but I'll probably copy some code for >> this from the smiles format. > > That makes a lot of sense. It definitely seems like trunk is getting much > closer to "bullet-proof" which is a good sign before a release. Next come > thousands of users who will test our definition of bullet-proof. :-) All 5 million structures now pass the shuffle test without stopping or segfaults. Tim > Thanks, > -Geoff ------------------------------------------------------------------------------ Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today. http://p.sf.net/sfu/beautyoftheweb _______________________________________________ OpenBabel-Devel mailing list OpenBabel-Devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/openbabel-devel
