Is it possible to create SMILES representation out of DNA/RNA-Oligonucleotides (~50 nts)?

Question

I want to use physico chemical descriptors of single stranded DNA/RNA-Oligonucleotides in order to model certain properties. To generate these descriptors i need a computer understandable representation of these oligonucleotides, e.g. the SMILES representation. So far i could not find of any method to generate these SMILES representations out of the primary structure of these oligonucleotides. I am wondering if someone knows of a software, that takes the primary sequence of a single stranded DNA-Oligonucleotide with e.g. 50 nts as input and gives the SMILES representation as output. So far i could find nothing that would accomplish this task, although, at least to me, it seems like a straight forward task.

I will appreciate every constructive thought on that.

Thanks!

EDIT: Add some form of input and output file

As requested here an example of input and output pair:

input should be primary sequence: TCA

output should be the corresponding SMILES string: O=C(C(C)=CN1[C@@]2([H])C([H])([H])C@@([H])C@([H])O2)NC1=O

This example was generated with chemdraw 20.1.1.

For my use case i need a software, that convert many (like 50k - 1000k) primary sequences of single stranded DNA/RNA with a length of about 50 nts into the corresponding SMILES string. Ideally would be some sort of python package but anything else would help also a lot!

Answer 1

It looks like various chemical software toolkits can export SMILES, for example OpenBabel, using the babel command line tool. I am not readily seeing the best way to read e.g. a FASTA, though. Another option would be ChemDraw.

This researchgate thread has a lot of information on peptide->SMILES conversion, I don't see nucleotides though.

One option, though it would be a little finicky, would be to look up the nucleotides SMILES strings on PubChem and then write a script to string them together (with a monophosphate) based on an input sequence. It seems like there is some related tooling already in rdkit, but a convenient API doesn't seem to exist yet. The developer does have one suggestion for peptides, which would translate into something like this for nucleotides:

from rdkit import Chem

nt_mol = Chem.MolFromFASTA("ACATATAT", flavor=9)  # flavor param notes as DNA
Chem.MolToSMILES(nt_mol)

To do this across many nucleotides you could hook it up to biopython to iterate across a FASTA file.

The nucleotide flavor option exists, but there is a note that only peptide is supported. I don't know whether that means the feature is experimental or is not implemented.