Re: [Cdk-devel] SMARTS question

[Andrew D. <da...@da...>]

On Jun 26, 2008, at 2:37 AM, Rajarshi Guha wrote:
> But would it be handy to have such a 'group-level' logical OR in  
> SMARTS itself? I realize that it'd be more syntactic sugar than a  
> real, new feature but it does seem useful.

   o  SMARTS are used in SMIRKS, but this SMARTS extension can't be

   o  In most toolkits, the first atom in the match object is from
       the first atom pattern in SMARTS, the second atom corresponds
       to the second, and so on.  This extension weakens that.

   o  Should there be any way to figure out which "group" matched?
       If I'm reading the CDK code correctly, the matching code only
       returns List<List<Integer>>, where the integers are atom  
indicies.

      Therefore, given
          SMILES: c1ccccc1  SMARTS: [c1ccccc1, cccccc]
      it's not possible to figure out if the first or the second
      subgroup was the one that matched, under the current API.

   o  How many matches are in the following?
        SMILES: CC    SMARTS: [[C,O]C,CC]

      What about "unique matches" ("umatch" in Daylight terms)?

      And what about in
        SMILES: CC    SMARTS: [CC,CC,CC,CC]


				Andrew
				da...@da...