namespace_mapping.py

import mygene

def namespace_mapping(names, map_from=['symbol', 'alias'], map_to='symbol', species='human'):

    names = pd.Series(names)

    print(f"passed {len(names)} symbols")

    names_stripped = names.str.strip()
    if any(names_stripped != names):
        print(f"{sum(names.str.strip() != names)} names contained whitespace. Stripping...")

    names_stripped_unique = names_stripped.unique()
    if len(names_stripped_unique) != len(names_stripped):
        print(f"{len(names_stripped) - len(names_stripped_unique)} duplicates. {len(names_stripped_unique)} uniques.")

    print()
    mg = mygene.MyGeneInfo()
    out, dup, missing = mg.querymany(names_stripped_unique.tolist(), scopes=map_from, fields=[map_to], species=species, as_dataframe=True, returnall=True).values()
    out = out.reset_index().rename(columns={'query':'input'}).sort_values(['input', '_score'], ascending=[True, False]).drop_duplicates(subset='input', keep='first')

    same = out[out.input == out[map_to]]
    updates = out[(out.input != out[map_to]) & (out.notfound.isna() if 'notfound' in out else True)].set_index('input')[map_to]

    print(f"\nunchanged: {len(same)};  updates: {len(updates)};  missing: {len(missing)}")

    names_updated = updates.reindex(names_stripped.values)
    names_updated = names_updated.fillna(names_updated.index.to_series()).values

    return updates, missing, names_updated