ecotaxa.py

from pathlib import Path

import pandas as pd

def write_ecotaxa_tsv(patches_csv_path: str, where: str, sample_id: str, scope_id: str):
    # import patch output table
    df_patches = pd.read_csv(patches_csv_path)
    df_patches['img_file_name'] = df_patches['patch_filename'].apply(lambda x: '2d_patches_annotation/' + x)

    # make second column index level to comply w/ EcoTaxa schema
    df_patches.columns = pd.MultiIndex.from_frame(
        pd.DataFrame([
            df_patches.columns,
            df_patches.dtypes.apply(
                lambda x: '[f]' if x in ['float64'] else '[t]'
            )
        ]).T,
        names=['variable', 'data type']
    )

    # add new columns for ecotaxa
    df_patches.loc[:, ('acq_instrument', '[t]')] = 'Other microscope'
    df_patches.loc[:, ('acq_instrument_microscope', '[t]')] = scope_id
    df_patches.loc[:, ('sample_id', '[t]')] = sample_id

    df_patches.loc[:, ('acq_id', '[t]')] = df_patches.loc[:, ('input_file', '[t]')]
    df_patches.loc[:, ('object_id', '[t]')] = df_patches.loc[:, ('patch_id', '[t]')]
    df_patches.loc[:, ('process_id', '[t]')] = df_patches.loc[:, ('patch_id', '[t]')]

    cols_to_transfer = [
        'img_file_name',
        'object_id',
        'acq_id',
        'acq_instrument',
        'acq_instrument_microscope',
        'sample_id',
        'process_id'
    ]
    df_export = df_patches.loc[:, pd.IndexSlice[cols_to_transfer, :]]
    df_export.to_csv(Path(where) / 'ecotaxa.tsv', sep='\t', index=False)