orthologs.py

from gprofiler import GProfiler
gp = GProfiler(return_dataframe=True)

def gprofiler_orthologs(query, human_to_mouse=False, mouse_to_human=False, organism='mmusculus', target='hsapiens', returnall=False):

    if isinstance(query, pd.Index): query = query.tolist()
    elif isinstance(query, pd.Series): query = query.values.tolist()

    q = [x for x in np.unique(query).tolist() if str(x) != 'nan']
    if len(q) != len(query): print(f'{len(q)} unique of {len(query)}')

    if human_to_mouse: organism='hsapiens'; target='mmusculus'
    elif mouse_to_human: organism='mmusculus'; target='hsapiens'

    df = gp.orth(organism=organism, query=q, target=target).replace('N/A', np.nan)

    print(df)

    missing = df.loc[df.ortholog_ensg.isnull(), 'incoming'].values
    print(f'{len(missing)} missing out of {len(query)}')
    multimappers = [gene for gene, b in (df.groupby('incoming').size() > 1).items() if b]
    print(f'{len(multimappers)} multi-mappers out of {len(query)}')

    if returnall: return df

    df = df.dropna(subset=['ortholog_ensg'])
    df = df[df.n_result == 1]
    df = df.set_index('incoming')[['converted', 'ortholog_ensg', 'name']].rename_axis('query')

    return df