SMILES strings for the polymers

[weitse-hsu]

I wonder if there are any SMILE strings available in PDBbind-Opt for the polymers.

I know that one can obtain the SMILES strings of the small-molecule ligands by simply getting the CCD codes from pdbbind_opt.csv and fetching the SMILES strings from RCSB. However, for polymers, I wonder if there are any SMILES strings readily available in PDBbind-Opt. If not, what would be the way you recommend for retrieving or generating SMILES strings? I imagine one could run Chem.MolToSmiles on the output from mol_from_seq in fix_polymer, which I assume should be the same as converting the provided SDF files using RDKit. However, I am not entirely sure if that is reliable since RDKit might not handle all cases correctly.

Thank you so much in advance!

[Ericwang6]

Hi @wehs7661 , thanks for contacting us. That's a good question. Yes you are right - the current safest way to get SMILES strings is to convert from the SDF files (*_ligand_refined.sdf). Using mol_to_seq function can only generate a correct reference SMILES string when the polymer is a non-cyclic peptide consisting common amino acids. Two cases cannot be handled by this function and they will not be included in the dataset:

• Non-peptide polymers, e.g. polynucleotide/polysaccharide. Although there is a feasible way to generate a SMILES through the sequence, we didn't include them in the current version. Therefore, the current workflow cannot generate a reference SMILES for such cases.
• Cyclic peptides. Since there SMILES string is generated based on the sequence by assuming the peptide is non-cyclic. These cases will also be filtered out because ligand fixing module is not able to handle cases where the bond connectivity is different between the reference structure and structure to be fixed.

I hope this answers your question.

[weitse-hsu]

Hi @Ericwang6, thanks so much for your prompt reply! That does answer my question.

I have a separate quick question though. (I'm happy to move this question to an individual issue if you prefer.) I noticed that some entries in pdbbind_opt_sm might have been misclassified. For example, 2k46 has a disaccharide (alpha-D-glucopyranose-(1-3)-alpha-D-glucopyranose) as the ligand and the name of its subfolder is 2k46_GLC-GLC_B_1-2, which does align with the naming convention for polymer ligands. However, it was placed in pdbbind_opt_sm instead of pdbbind_opt_poly.

Thanks again!