CSEARCH- censo and cregen functionality

[hklem]

Hello! Really enjoying the AQME package so far. My colleagues and I have been using the CSEARCH (--program=crest) module to generate ensembles. We then use CENSO to narrow down the number of conformers, and a final cregen sorting of the CENSO results to get the conformers we will take forward to higher level QM calculations using QPREP and QCORR .
We are wondering if there is interest in adapting CSEARCH to 1) call CENSO and also 2) cregen without a crest conformer search? We would love to be able to use AQME for the whole workflow. Out initial input is a CSV containing the smiles strings of a big data set. It is extremely helpful that CSEARCH program=crest generates the ensemble SDF files that contain charge and multiplicity information, since those values vary in our data set. But, CENSO (we think) does not work with SDF, so we have been using obabel to convert file types and then we loose the charge and multiplicity information. Then, again when we do a final cregen sorting of the CENSO output (XYZ format), out output from that is XYZ as well. So right now we are considering doing the following procedure: 1. CSEARCH program=crest, 2) CENSO, 3) crest --cregen, 4) QPREP; which will require us to add additional steps to bring back the charge and multiplicity information when we call QPREP. But we would like to propose this procedure: 1. CSEARCH program=crest, 2) CSEARCH program=censo (input is *sdf, the crest ensembles of each molecule), 3) CSEARCH program=crest (but doesn't run crest search, only calls cregen from crest to do a final sorting), and the output would be the clustered ensembles in SDF format (like the output from step 1) 4) QPREP (input is *sdf)