unexpected ligand pka values

[mbatgh]

Hi,

I just used propka3 (as downloaded and installed from github) to analyze a few structures from the PDB.
I used the "-o 7" command line option, and other than that, all defaults.
The results look in most cases as expected, but I have three cases were the results for bound ligands are very unexpected.
these are: 7JTO, 7JTP, 6W8I (the output .pka files from propka are attached)
what I see is:
7JTO: I get pka=10.66 for a ligand nitrogen (chain B res VKA N65) which is directly bound to an aromatic ring
7JTP: as above for X6M N1, also there is one more piperazine N (chain A res X6M N) which propka seems to ignore.
6W8I: chain D res TKY N50, pka = 9.65 - this is an amide nitrogen

If I understood this correctly then, at pH~7, all the nitrogens mentioned aboved should be protonated/charged according to propka, which I cannot see happening for these compounds.
Please let me know in case I overlooked/misunderstood something here!

thanks,
Michael

pkaoutputs.zip

[jhjensen2]

For some reason the code is assigning the wrong atom type to these Ns. For example for 6W8I N50 is assigned to be a tertiary N (N33), but should be an amide (NAM) and thus non-tritrateable.

So these problems appear to be due to bugs in the code. I am no longer active on PropKa, but perhaps someone can look into this?

[mbatgh]

Thanks for the reply ... I take it the relevant code is supposed to be in: ligand.py
I was, in fact surprised, that propka tries to infer atom types/bond orders, etc from
PDB input alone, as this is rather difficult and error prone, is there a way to supply an sdf or mol2 file
instead of relying on the PDB?