liana.rs.translate_column

liana.rs.translate_column(resource, map_df, column, replace=True, one_to_many=1)

Generate orthologs for a given column in a DataFrame.

Parameters:

resource DataFrame

Input DataFrame.

map_df DataFrame

DataFrame with orthology mappings, where the first column is the source and the second column is the target for mapping.

column str

Column name to translate.

replace bool (default: True)

Whether to replace the original column with the translated values. Default is True. If False, it will create a new column with the prefix orthology_.

one_to_many int (default: 1)

Maximum number of orthologs allowed per gene. Default is 1.

Notes

This function generates orthologs for a given column in a DataFrame. It handles complex names by splitting them into subunits and generating all possible combinations of orthologs. It assumes that subunits are separated by an underscore (“_”).

Return type:

DataFrame

Returns:

Resulting DataFrame with translated column.

Raises:

ValueError – If the mapping_df does not contain ‘source’ and ‘target’ columns or one_to_many is not an integer

Examples

map_df maps human symbols (source) to the target organism (target). liana.rs.get_hcop_orthologs() builds one; it is written out here to keep the example offline:

>>> import pandas as pd
>>> import liana as li
>>> resource = li.rs.select_resource("consensus").head(3)
>>> map_df = pd.DataFrame(
...     {"source": ["LGALS9", "PTPRC", "MET", "CD44"], "target": ["Lgals9", "Ptprc", "Met", "Cd44"]}
... )
>>> li.rs.translate_column(resource, map_df, column="ligand")
   ligand receptor
0  Lgals9    PTPRC
1  Lgals9      MET
2  Lgals9     CD44

With replace=False the translation is added as an orthology_ligand column instead of overwriting ligand. Use liana.rs.translate_resource() to do both sides at once.