Noel, On Tue, Aug 5, 2008 at 3:49 PM, Noel O'Boyle <[email protected]> wrote: > For purposes of timing comparison (C++ API vs. using Boost.Python), is > there any operation available from the C++ API that takes about 0.5 > seconds per molecule? I've tried to slow down the Forcefield > Optimization by reducing the tolerances but it's much too fast still.
Heh... it's not often one hears the "complaint" that their code is too fast. Nice flattery there Noel. :-) Probably the easiest thing you can do to find an operation that takes about 0.5 seconds per molecule is to use the distance-geometry based embedding code and play around with the number of conformations per molecule until you're happy with the time required. I'm sure you'll be able to get the time to 0.5 seconds per molecule. The function you want is RDKit::DGeomHelpers::EmbedMultipleConfs, which is documented here: http://tinyurl.com/6fmxa5 -greg
