I have an SDF file with about 100 molecules, and 100 corresponding experimentally determined pKa's. When training the pKa plugin via cxtrain, how should I assign the atom numbering?
If I open the SDF directly in MarvinView and display atom numbers, I get a different atom numbering than if I copy a molecule's SMILES from the SD-file and paste it into MarvinSketch and show atom numbering.
There should be no difference between the atom numbering, regardless of the format used for file manipulations. Can you please send us your SD file so that we can reproduce if there is a format bug?
Meanwhile I would send you our pKa training manual, where you can find detaild description of how to train our pKa calculation.