MechanophoresResearch/AutoencoderDataset

SMILES Splitter Script

The file splitter script (split_file.sh) is used to split large SMILES files into smaller parts.

Usage

./split_smi.sh <input_file> [max_size_mb]

• <input_file>: The SMILES file to split
• [max_size_mb]: Optional. The maximum size of each part in MB (default: 1024)

Running with xargs for parallel processing

To split multiple files in parallel:

find . -name "*.smi" | xargs -n1 -P$(nproc) ./split_file.sh

To specify a custom size (e.g., 500 MB):

find . -name "*.smi" | xargs -n2 -P$(nproc) ./split_file.sh {} 500

Note: This script will delete the original file if it was split successfully.

SELFIES Converter Script

The SELFIES converter script (selfies_converter.py) converts SMILES files to SELFIES format.

Usage

python selfies_converter.py <input_file> [-v] [-b BATCH_SIZE]

• <input_file>: The SMILES file to convert
• -v, --verbose: Optional. Increase output verbosity
• -b BATCH_SIZE, --batch-size BATCH_SIZE: Optional. Batch size for processing (default: 10000)

Running with find and xargs for parallel processing

To convert all .smi files in parallel:

find . -name "*.smi" | xargs -I {} -P$(nproc) sh -c 'python selfies_converter.py {}'

To convert files and remove the original .smi files after successful conversion:

find . -name "*.smi" | xargs -I {} -P$(nproc)  sh -c 'python selfies_converter.py {} && rm {}'

Note: Be cautious when using the command that removes original files. Ensure you have backups before running it.

With gz:

find . -name "*.smi.gz" | xargs -P$(nproc) -I {} sh -c 'f="{}"; g="${f%.gz}"; gzip -dc "$f" > "$g" && python selfies_converter.py "$g" && rm "$g" && rm "$f"'

Output

Both scripts will create output files in the same directory as the input files:

• The file splitter creates <original_name>.part01.smi, <original_name>.part02.smi, etc.
• The SELFIES converter creates <original_name>.selfies for the converted file and <original_name>.alph for the alphabet file.