utils/metadata.py

import pandas as pd

def read_metadta(file_path):
    last_6_columns = [
        "Data creation year",
        "Data creation month",
        "Data creation day",
        "Specimen collection year",
        "Specimen collection month",
        "Specimen collection day"
    ]

    # Create a dtype dictionary to set the last 6 columns as strings
    dtype_dict = {col: str for col in last_6_columns}

    metadata_df = pd.read_csv(file_path, dtype=dtype_dict)
    return metadata_df