Monday, 21 March 2011

Confab, a systematic conformation generator, published in J Cheminf

My paper on Confab has just been published by the Journal of Cheminformatics. Confab is a systematic (and exhaustive) conformation generator that uses a torsion-driving approach to generate all diverse (by RMSD) low-energy conformers.

Confab - Systematic generation of diverse low-energy conformers NM O'Boyle, T Vandermeersch, CJ Flynn, AR Maguire, GR Hutchison. Journal of Cheminformatics 2011, 3:8.

The software itself is available from http://confab.googlecode.com.

Here's the abstract of the paper:
Background
Many computational chemistry analyses require the generation of conformers, either on-the-fly, or in advance. We present Confab, an open source command-line application for the systematic generation of low-energy conformers according to a diversity criterion.

Results
Confab generates conformations using the 'torsion driving approach' which involves iterating systematically through a set of allowed torsion angles for each rotatable bond. Energy is assessed using the MMFF94 forcefield. Diversity is measured using the heavy-atom root-mean-square deviation (RMSD) relative to conformers already stored. We investigated the recovery of crystal structures for a dataset of 1000 ligands from the Protein Data Bank with fewer than 1 million conformations. Confab can recover 97% of the molecules to within 1.5 A at a diversity level of 1.5 A and an energy cutoff of 50 kcal/mol.

Conclusions
Confab is available from http://confab.googlecode.com.

11 comments:

Jan said...

Is that the approach that Avogadro uses to search for conformers? If not, is it going into Avogadro?

baoilleach said...

I guess that the systematic search is pretty much the same. Confab handles symmetry a bit better - maybe this could be merged with Open Babel/Avogadro (I'd have to check).

chembioinfo said...

Sorry for being naive. So from the protocol as described in your paper, the starting point is a pre generated 3D strictures?

a) If yes, then does the quality of the 3D structures matter (in terms of energy landscape)?

b) Will it work on 2D too (will the tool automatically generate 3D conformers from 2D and then use protocol)?

baoilleach said...

@chembioinfo: Yes, that's right.

Regarding (a), I would say no in general, but all of the conformers will have higher MMFF94 energies than if they had been optimised with the MMFF94 forcefield in advance. Relative to each other, there shouldn't be significant differences (although there will be some).

It won't work with 2D too. 2D->3D is a problem all on its own, and I decided it was better to separate them out here. (This is not uncommon, although it would be nice to support 0D/2D input too.)

chembioinfo said...

Thanks, nice work!

Scott said...

Looks great, I had two questions though.
(1) To get the maximum number of conformers you should use minimal RMSD cutoff and maximum cutoff energy? And this would give the most detailed representation of the energy landscape?
(2) To find the best conformer, you should do a MMFF94 optimization for each and then select the lowest energy result?

Cheers, Scott

baoilleach said...

@Scott:
(1) Yes. You should also max out on the number of conformers generated.
(2) Well, I guess so if you want the the energy landscape in terms of MMFF94. The only problem is the time required to do this. Obviously, there is no guarantee that the absolute minimum according to MMFF94 is the desired conformation.

Scott said...

Many thanks. The general goal is to go from smiles, generate 3d coords with openbabel (gen3d) and then to find an 'accurate' conformer by using confab and MMFF94 (or other). The resulting coordinates will then be used for a gaussian dft calculation. I guess the main issue is how long to spend on the conformer searching. I'd be very interested to hear your thoughts on this type of approach

anonymous said...

Is there any free conformer generation program considering ring flexibility?

That would be the logical first step prior to running Confab.

Noel O'Boyle said...

I don't know of any.

anonymous said...

Balloon, for one, can take into account ring flexibility. Also DG-AMMOS and some other free codes. The problem is that I could not find a way to filter the results according to the RMSD of atoms belonging to rings.

If one could do that, it would be possible to, for instance, generate 100 conformers with Balloon, keep only those with a significantly different ring conformations and then feed Confab with those...