SMILES Parser and Validator

Parse, validate, lightly clean, and summarise SMILES strings using RDKit.

The script can read SMILES directly from the command line or from a text file, then writes a CSV report containing parsing status, canonical SMILES, formula, molecular weight, InChIKey, and atom counts.

Features

• Accepts SMILES from command-line arguments or a plain text file.
• Validates structures with RDKit.
• Optionally applies a simple cleanup pass for noisy explicit-hydrogen strings.
• Calculates useful molecular properties.
• Writes a structured CSV report with one row per input SMILES.

Requirements

• Python 3.8+
• RDKit

Recommended installation with conda:

conda install -c conda-forge rdkit

Input Options

Command-line SMILES

Pass one or more SMILES strings with --smiles:

python parse_smiles.py --smiles "CCO" "O=C(O)C" -o results.csv

Text File Input

Use --infile for a text file containing one SMILES string per line.

An example file, my_smiles.txt, is included in this folder and can be used for a quick test:

python parse_smiles.py --infile my_smiles.txt --out smiles_parsed.csv

Cleaning Behavior

By default, the script tries a simple cleanup pass if RDKit cannot parse the original SMILES.

The cleanup is intentionally basic and targets noisy explicit hydrogen patterns such as:

• O[H]
• [H]
• Empty parentheses left after removing [H]

This can help with exported or messy SMILES, but it may alter the intended chemistry. To disable this behavior:

python parse_smiles.py --infile my_smiles.txt --out smiles_parsed.csv --no-clean

Output Columns

The output CSV contains:

• input_smiles: original input string
• used_smiles: original or cleaned string used for successful parsing
• status: parsing result
• error: error details when parsing fails
• canonical_smiles
• isomeric_smiles
• formula
• mol_wt
• exact_mw
• inchi_key
• num_atoms
• num_heavy_atoms

Example Workflow

python parse_smiles.py --infile my_smiles.txt --out smiles_parsed.csv

Expected terminal output:

Wrote 7 rows to smiles_parsed.csv

Open smiles_parsed.csv to inspect which structures parsed successfully and which, if any, failed.