Some problems about the SMILES about caster HOT 5 CLOSED

Thanks, the problems have been solved.
It seems that you're invited by Prof. Zeng for a report tomorrow, and I'm looking forward to attending it.

from caster.

Hi Yifan,

The chembl_seq.txt is a txt file containing all SMILES in the ChEMBL dataset. Each line is a SMILES string.

Yeah, the SMILES strings should be canonicalized. You can achieve this by "Chem.MolToSmiles(Chem.MolFromSmiles(SMILES_STRING))"

I suggest you generate your own ESPF fingerprint for the best performance after canonicalization. To generate, you can checkout this notebook example: https://github.com/kexinhuang12345/ESPF/blob/master/generate_drug.ipynb

from caster.

Hi Kexin,

The problems have been solved. However, there is still another problem which bothers me a lot. For example, for Drug:Cisplatin (DB00515), the SMILES from Drugbank is [H]N([H])Pt(Cl)N([H])[H]. The Chem.MolFromSmiles didn't work for this SMILES. (Similar circumstances occur for other SMILES which contain [Pt]). In order to deal with it, I downloaded the Mol files for these drugs and use Chem.MolToSmiles(Chem.MolFromMolFile("DB00515.mol")), and get the result of 'NPt(Cl)Cl'.

Is the procedure right?

from caster.

Yes, the issue is the element Pt should be wrapper like [Pt]; if you try [H]N([H])PtN([H])[H]; it should work.

The procedure you described also works too; 'NPt(Cl)Cl' and '[H]N([H])Pt(Cl)N([H])[H]' are the same thing, since I guess the default Chem.MolFromMolFile() may automatically remove the hydrogen key in the SMILES.

from caster.

Hi!
Thanks for your work.
Giving "chembl_seq.txt" along with the code, will be very helpful for people like me that have just started studying and doesn't have a clear clue on how to obtain it, but still want to use your code and be able to generate his own ESPF fingerprint.
Can you give it now, please?

from caster.