From: Egon W. <ego...@gm...> - 2007-06-26 05:17:38
|
On 6/25/07, Rajarshi Guha <rg...@in...> wrote: > On Jun 25, 2007, at 4:09 PM, Egon Willighagen wrote: > > The method IMolecularDescriptor.getDescriptorType() sounds applicable > > to all descriptors. Should i not move to IDescriptor? > > Sounds reasonable. OK, will file this as RFE. > BTW, I was trying to update SmilesParser.java at line 684 - checking > for UNSET before using the hydrogen count. This will clean up some of > the failing unit tests. Good. BTW, got a 404 on your home directory! > However after making the fix, some of the unit tests (I think in the > smiles module) seem to run forever. Was Nina's patch for > interruptible SMILES parsing included? Yes, think so. I'll use this JUnit timeout feature, and will set it to 1 second. Any SMILES should really parse well within that limit. I think the most time consuming step in SMILES parsing is currently really the deduction of bond orders, which I really would like to propose not to do, and to simply regards c1ccccc1 systems as they are in the file format: sp2 carbons. From that we can still do simple aromaticity detection based on sp2 only ring systems, and etc... Will make some time this week to process SMILES from PubChem and compare them to the corresponding InChIs... watch my blog. Egon |