import re
def mitab_parser(mitab:str, uniprot_ac:list[str])->list[ list[str] ]:
""" Extract from provided file (path), lines featurin at least one interactor
present in uniprot_ac provided list
Returns: A list of mitab records split over tabulation.
"""
hits = []
with open(mitab, 'r') as fp:
for line in fp:
line = line.split()
for proteinfield in line[0:2]: # Test column One and Two
_ = re.match("uniprotkb:(\\S+)", proteinfield) # Extract uniprot identifier
x = _[1] if _ else None
if x in uniprot_ac: # Check uniprot identifier is in provided ac list
hits.append(line)
break
print(f"Filtered {len(hits)} interactions")
return hits
res = mitab_parser("/Users/glaunay/t.txt", ['P0A6F5', 'P02931'])import pandas
df = pandas.read_csv('data/TCL_wt1.tsv', sep="\t", na_values="#VALEUR!")
df = df.dropna()
df_abd = df.loc[(df['-LOG10 Adj.P-val'] > 3 ) &
(df['Log2 Corrected Abundance Ratio'] > -0.6 ) ]
abd_acc = df_abd['Accession'].tolist()GO_dic = {}
for acc in abd_acc:
go_terms = getAccessionGOTerms(
"./data/uniprot-proteome_UP000000625.xml", acc)
for go_id, go_name in go_terms:
if not go_id in GO_dic:
GO_dic[go_id] = {
'ID' : go_id, 'name' : go_name, 'carried_by' : []
}
GO_dic[go_id]['carried_by'].append(acc)# Background statistics
import json
from scipy.stats import hypergeom
background = {}
with open("data/EColiK12_GOcounts.json", "r") as fp:
background = json.load(fp)["go_terms"]
"""eg:
{'GO:0000006': {'count': 1,
'name': 'F:high-affinity zinc transmembrane transporter activity'}
"""
def compute_pvalue(GO_term:dict, background, ttl_abdnt :int, N = 1800):
""" Compute pvalue of enrichissment of passed GO_term
Where GO_term is of the shape :
{ 'ID' : 'GO:0000006', 'name' : "blabla", 'carried_by' : ["P0000", "P11111"] }
"""
GO_id = GO_term['ID']
GO_name = GO_term['name']
n = ttl_abdnt # Total nb of draw
k_obs = len(GO_term["carried_by"]) # observed nb of success
K = background[GO_id]['count']
rv = hypergeom(N, K, n)
p_value = 0 # pvalue = P(X>=k_obs)
for k in range(k_obs, ttl_abdnt + 1):
p_value += rv.pmf(k)
return p_value, GO_id, GO_name
compute_pvalue(GO_dic['GO:0000062'], background, 23, 1800)
# Applied to all GO go_terms
all_scores = []
for GO_ID in GO_dic:
all_scores.append(
compute_pvalue(GO_dic[GO_ID], background, 23, 1800)
)
sorted(all_scores, key = lambda t:t[0])