Re: [Open Babel] OBMol2Smi conversion speed

[Geoff H. <ge...@ge...>]

On Dec 2, 2004, at 3:02 AM, Daen de Leon wrote:

> Could somebody tell me what it is about the SMILES conversion 
> calculations that makes them so CPU intensive, and is there any 
> feasible way of doing some speedup?  I haven't tried other output 
> formats yet.  Perhaps MOL files would be quicker?

I have not done extensive performance profiling--especially on the 
number of compounds you're using. So I can speak for sure. But yes, I 
suspect SDF Mol or Mol2 files would be faster.

The SMILES format requires quite a bit of "extra" information. For 
example, you must find ring structures, assign cis/trans 
stereochemistry, etc. Using the "lazy evaluation" in Open Babel, the 
routines called to perceive this data are not used in many other 
formats.

If you would be kind enough to file a bug report with as much 
information as you can provide, will be able to do additional 
performance profiling to see which routines in particular are called 
the most. (It would be nice to have some files with that many 
structures, though I suspect you don't want to pass your particular 
molecules along.)

http://sourceforge.net/tracker/?atid=428740&group_id=40728&func=browse

Cheers,
-Geoff