Fishers exact test from Google sheet columns

fishers_exact_test.py

from scipy.stats import fisher_exact
import pandas as pd
from IPython.display import display, Markdown

def read_google_sheet():
    SHEET_ID = '1JSjSLuto3jqdEnnG7JqzeC_1pUZw76n7XueVAYrUOpk'
    SHEET_NAME = 'Sheet1'
    url = f'https://docs.google.com/spreadsheets/d/{SHEET_ID}/gviz/tq?tqx=out:csv&sheet={SHEET_NAME}'
    df = pd.read_csv(url, header=1, dtype='str')
    return df.loc[:, [not x.startswith('Unnamed') for x in df.columns]]
    
def gene_list_names():
    df = read_google_sheet()
    return sorted(df.columns.tolist())

def gene_list(name):
    df = read_google_sheet()
    sr = df[name]
    return sr[~sr.isnull()]

def fisher(a, b, background=(None, None), alternative='greater', format_case=False):
    
    ((a_name, a_set), (b_name, b_set))  = a, b

    a_set = set(a_set)
    b_set = set(b_set)

    if background == (None, None):
        background_set = a_set.union(b_set)
    else:
        background_name, background_set = background
        background_set = set(background_set)
    
    M = len(background_set)
    N = len(background_set.intersection(a_set))
    n = len(background_set.intersection(b_set))
    x = len(background_set.intersection(a_set).intersection(b_set))

    table = [[  x,           n - x          ],
             [ N - x,        M - (n + N) + x]]

    pvalue = fisher_exact(table, alternative=alternative).pvalue
    
    if background != (None, None):
        display(Markdown(f"Among {len(background_set)} **{background_name}** genes:"))  
    if format_case:
        display(pd.DataFrame().from_records(table, 
                                    columns=[a_name.isupper() and a_name or a_name.capitalize(), 
                                             f'Not {a_name.isupper() and a_name or a_name.lower()}'],
                                    index=[b_name.isupper() and b_name or b_name.capitalize(), 
                                           f'Not {b_name.isupper() and b_name or b_name.lower()}']
                                   ))
    else:
        display(pd.DataFrame().from_records(table, 
                                    columns=[a_name, f'not {a_name}'],
                                    index=[b_name, f'not {b_name}']
                                   ))

    display(Markdown(f'**{a_name}** x **{b_name}** overlap has **p-value:** {pvalue:.2e}'))
    text = ', '.join(sorted(background_set.intersection(a_set).intersection(b_set)))

    display(Markdown('**overlap:**<br>  ' + text))

asd_set = set(gene_list('sfari_all_conf'))
spermatid_set = set(gene_list('meritxell_spermatid_expr'))
brain_set = set(gene_list('hpa_brain_prot'))

fisher(('Brain', brain_set), ('spermatid', spermatid_set), background=('ASD', asd_set))