def save_enrichment(x):
lib = gp.get_library_name('Human')
with open('gensets.txt', 'w') as f:
for item in range(len(lib)):
f.write("%s %s\n" % (item, lib[item]))
# lib = lib[49: 54]
lib = lib[53]
files = [(1, x+"/gcn-hom-hom.csv"), (2, x+"/gcn-hom-onto.csv"),
(3, x+"/gcn-onto-onto.csv"), (4, x+"/gae-hom-hom.csv"),
(5, x+"/gae-hom-onto.csv"), (6, x+"/gae-onto-onto.csv")]
df = pd.DataFrame()
writer = pd.ExcelWriter('enrich-cluster/full-results.xlsx')
for key, file in files:
print(file)
cluster_data = read_file_2(file)
for i in cluster_data:
try:
enr = gp.enrichr(gene_list=list(cluster_data[i][2]), gene_sets=lib, organism='Human', cutoff=0.05).results
except:
pass
enr['model'] = key
enr['cluster'] = i
df = df.append(enr)
df = df[(df['P-value'] < 0.05)]
df.to_excel(writer, sheet_name="sheet1")
writer.save()
def scatter_stats(perms):
statistics = {"gae-hom-hom-gae-hom-onto": list(),
"gae-hom-hom-gae-onto-onto": list(),
"gae-onto-onto-jcd-onto": list(),
"gae-hom-hom-jcd-hom": list(),
"gae-hom-hom-jcd-onto": list(),
"gae-hom-onto-gae-onto-onto": list(),
"gae-hom-onto-jcd-hom": list(),
"gae-hom-onto-jcd-onto": list(),
"gae-onto-onto-jcd-hom": list(),
"jcd-hom-jcd-onto": list(),
}
for perm in range(perms):
print("********** permutation no {} *********".format(perm))
random_elments = random.sample(list(read_file_2("../enrich_red/selected_genesets.csv").keys()), 50)
for i in files:
print("file is {}".format(i))
file_one = "../enrich_red/" + i[0] + ".csv"
file_two = "../enrich_red/" + i[1] + ".csv"
_one = read_file_2(file_one)
_two = read_file_2(file_two)
one = {ii: _one[ii] for ii in random_elments}
two = {ii: _two[ii] for ii in random_elments}
array = []
for x, y in zip(one, two):
x_one = list(one[x])
y_one = list(two[y])
array.extend(res(x_one, y_one))
array.extend(res(y_one, x_one))
x, y = zip(*array)
linreg = spy.stats.linregress(x, y)
statistics[i[0]+"-"+i[1]].append(linreg.rvalue)
write_file("../perms2/"+lib+".csv", statistics)
def one_correlation():
selected = read_file_2("../enrich_red/selected_genesets.csv")
top_n = 20
for i in files:
similar = read_file_2("../enrich_red/" + i + ".csv")
array = []
for x, y in zip(selected, similar):
x_one = list(selected[x])
y_one = list(similar[y])
array.extend(res(x_one, y_one, top_n))
x, y = zip(*array)
fig, ax = plt.subplots()
ax.scatter(x, y)
linreg = spy.stats.linregress(x, y)
temp = [linreg.intercept + linreg.slope.item() * k for k in x]
plt.plot(x, temp, 'r')
x_min = min(x)
x_max = max(x) + statistics.stdev(x)
y_min = min(y)
y_max = max(y) + statistics.stdev(y)
plt.xlim(x_min, x_max)
plt.ylim(y_min, y_max)
plt.ylabel("Corresponding P-value")
plt.xlabel("Actual P-value")
# plt.text(i[2], i[3], 'R2 = %0.2f' % linreg.rvalue)
# plt.text(i[2], i[4], 'Slope = %0.2f' % linreg.slope)
extra = Rectangle((0, 0),
1,
1,
fc="w",
fill=False,
edgecolor='none',
linewidth=0)
ax.legend(
[extra, extra],
('R2 = %0.2f' % linreg.rvalue, 'Slope = %0.2f' % linreg.slope))
plt.title("Correlation plot for " + i)
plt.show()
def stat_correlation(perms, top_n):
statistics = {
"gae-hom-hom": list(),
"gae-hom-onto": list(),
"gae-onto-onto": list(),
"jcd-hom": list(),
"jcd-onto": list()
}
for perm in range(perms):
print("********** permutation no {} *********".format(perm))
# random_elments = random.sample(list(read_file_2("ms-project/geneset_pairing.csv").keys()), 50)
random_elments = random.sample(
list(read_file_2("../enrich_red/selected_genesets.csv").keys()),
50)
_selected = read_file_2("../enrich_red/selected_genesets.csv")
selected = {ii: _selected[ii] for ii in random_elments}
for i in files:
print("file is {}".format(i))
_similar = read_file_2("../enrich_red/" + i + ".csv")
similar = {ii: _similar[ii] for ii in random_elments}
array = []
for x, y in zip(selected, similar):
x_one = list(selected[x])
y_one = list(similar[y])
array.extend(res(x_one, y_one, top_n))
x, y = zip(*array)
linreg = spy.stats.linregress(x, y)
statistics[i].append(linreg.rvalue)
write_file("../perms/" + lib + ".csv", statistics)
def distribution():
mthds = read_file_2("../perms2/GO_Biological_Process_2018.csv")
# Using 95% confidence interval
# (1-0.95)/2
t_score = abs(t.ppf(0.025, 23))
alpha = 1 - 0.95
excel_rows = [['Method', 'PT. Est', 'lower CI', 'upper CI']]
for i in mthds:
mean = statistics.mean(mthds[i])
std = statistics.stdev(mthds[i])
sqtr_nu = math.sqrt(len(mthds[i]))
# p_hat and q_hat set to conservative since we have no previous data #0.5 for each
# Since its probability I clip to 0
x = pd.Series(mthds[i])
# if i == 'jcd-hom-jcd-onto':
# mtd = replacenth(i, "-", " vs ", 2)
# else:
# mtd = replacenth(i, "-", " vs ", 3)
mtd = i
lower_ci = max(mean - t_score * std / sqtr_nu, 0)
upper_ci = mean + t_score * std / sqtr_nu
qq_plot(x, mtd, "../qqplots/qq-" + i + ".png")
excel_rows.append(
[mtd, round(mean, 3),
round(lower_ci, 3),
round(upper_ci, 3)])
df = pd.DataFrame.from_records(excel_rows[1:], columns=excel_rows[0])
print(df)
print(df.to_latex(index=True))
sample = read_file("enrich_red/gae-hom-hom.csv")
red = read_file("enrich_red/selected_genesets.csv")
temp = {}
for i in red:
if i in sample:
temp[i] = red[i]
write_file("enrich_red/selected_genesets.csv", temp)
# reduce_genesets()
lib = gp.get_library_name('Human')[53]
files = [("gae-hom-hom", 1, 1), ("gae-hom-onto", 1, 2),
("gae-onto-onto", 2, 3), ("jcd-hom-hom", 1, 4),
("jcd-onto-onto", 2, 5)]
data_desc = read_file_2("data\ms-project\data-description.csv")
for i in files:
file_name = "enrich_red/" + i[0] + ".csv"
file = read_file(file_name)
# if i[1] == 1:
# neigh = read_file_2("data/ms-project/neig_len_hom.csv")
# else:
# neigh = read_file_2("data/ms-project/neig_len_onto.csv")
if i[2] == 1:
rank = read_file_2(
"ranking_results/ -- GAE -- Homology -- Homology.csv")
elif i[2] == 2:
rank = read_file_2(
"ranking_results/ -- GAE -- Homology -- Ontology.csv")