There was recently a paper citing RDKit conformer generation as being "good" - see http://dx.doi.org/10.1021/ci2004658. It would be really nice to have the option to generate multiple (diverse) conformers within the co-ordinate generation node (or a new node), along with the calculated conformer energy - and perhaps options to control / limit the min and max no of conformers for each molecule.
I second that. That would be really useful.
This is a nice idea for a new node.
I guess the correct behavior would be to add a separate molecule to the output table for each conformer generated along with some kind of conformer ID?
One bit of early expectation management: the conformation energy in the RDKit is currently calculated using UFF, this is not a particularly good force field for energy calculations, so the energies will be extremely rough.
Perhaps a node for obconformer in openbabel since it uses mmff94 and openbabel conversion node is already present in knime?
I think that would be the ideal scenario. I guess if the energy is an issue, then the energies canbe calculated by alternative means from the conformers. Also, is it possible in RDKit to align the co-ordinates to the principle axes?