From: Uli F. <u.f...@ch...> - 2005-11-22 18:53:35
|
I actually experienced similar problems when I started using CDK. I would prefer the automatic configuration of molecules by Readers. However, there are file formats that allow the explicit specifiation of certain atom properties like e.g. an isotope. This should be taken into account: - if there is something specified in the file do not change anything; especially do not validitate in any way if the specified stuff is meaningful - if nothing is specified in the input file configure isotopes, hydrogens etc. automatically In other words: I want to emphasize that if something is explicitely specified in a file do not touch it. Uli Christoph Steinbeck wrote: > Hi everybody, > > we just had a discussion here with Peter MR and coworkers on how some > things in CDK are done. The starting point was a failing canonical > SMILES generation, which was failing because the respective molecule > a) was not configured by an isotope factory > b) had no implicit H's added by HydrogenAdder. > > Peter argued that this was less then obvious (where he is right) and > that the CDK.Readers, etc, should really spit out properly configured > molecules. > > What does everyone think? > Should our io.Readers do that or should we leave it like it is, i.e. do > only the minimum and let users plug things together. In the latter case, > things need at least to be documented. > > Cheers, > > Chris > |