PyOVERCAST 1.0.8__tar.gz → 1.0.9__tar.gz

{pyovercast-1.0.8 → pyovercast-1.0.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: PyOVERCAST
-Version: 1.0.8
+Version: 1.0.9
 Summary: A Python package for mining key transcription factors from transcriptome data.
 Author-email: Tinghua Huang <thua45@126.com>
 License-Expression: MIT
@@ -78,6 +78,9 @@ if __name__ == '__main__':
 	# or predict one DEG-list with bootstrap
 	result = predict.olcr_bootstrap(set_names=['human_jaspar_CLink_rci4_1w_0.8', 'human_hocomoco_CLink_rci4_1w_0.8'], list_file='./OVERCAST_data/input_deg-list.txt', win=30, bs_n=1000, thread_n=32)
 
+	# or predict one DEG-list with bootstrap and permutation
+	result = predict.olcr_bootstrap_permutation(set_names=['human_jaspar_CLink_rci4_1w_0.8', 'human_hocomoco_CLink_rci4_1w_0.8'], list_file='./OVERCAST_data/input_deg-list.txt', win=30, bs_n=1000, thread_n=32)
+
 	# save result to text file
 	result.to_csv('output.txt', sep='\t', index=False, encoding='utf-8')
 	

{pyovercast-1.0.8 → pyovercast-1.0.9}/PyOVERCAST/predict.py

@@ -13,8 +13,8 @@ plt.rcParams['font.sans-serif'] = ['SimHei', 'Microsoft YaHei', 'PingFang SC', '
 plt.rcParams['axes.unicode_minus'] = False  # 解决负号显示问题
 import multiprocessing as mp
 import seaborn as sns
-
 from scipy import stats
+import random
 
 overcast_data_dir = './OVERCAST_data'
 
@@ -520,6 +520,36 @@ def rrho_coor_bs(df1, df2, win, bs_n):
 
 	return beta_xy, pval_xy, DTCI, bs_pval
 
+def permutation(gene_set, gene_list):
+	random.seed(42)
+	gene_set0 = gene_set[gene_set.index.isin(gene_list.index)]
+	gene_list0 = gene_list[gene_list.index.isin(gene_set0.index)]
+
+	diff_idx = list(gene_list.index.difference(gene_list0.index))
+	if len(diff_idx) == 0:
+		gene_sample = []
+	else:
+		gene_sample = random.sample(diff_idx, gene_set0.shape[0])
+	
+	set1_values = []
+	set0_max, set0_min = gene_set0[gene_set0.columns[1]].max(), gene_set0[gene_set0.columns[1]].min()
+	set0_range = set0_max - set0_min
+	if set0_range < 0.0:
+		set0_range = set0_min
+	for gene in gene_sample:
+		set1_values.append(random.uniform(set0_min - set0_range, set0_min))
+	
+	if len(gene_sample) == 0:
+		gene_set_pt = gene_set0
+		gene_list_pt = gene_list0
+	else:
+		gene_set1 = pd.DataFrame({"gene": gene_sample, "logFC": set1_values}, index=gene_sample)
+		gene_set_pt = pd.concat([gene_set0, gene_set1], axis=0)
+		gene_list1 = gene_list[gene_list.index.isin(gene_sample)]
+		gene_list_pt = pd.concat([gene_list0, gene_list1], axis=0)
+
+	return gene_set_pt, gene_list_pt
+
 def thread_one(tfbs_array, deg_list, win, shared_list, nn, lock):
 	tf_n = len(tfbs_array)
 	epsilon = 1e-15   # 常用值
@@ -546,6 +576,19 @@ def thread_one_bs(tfbs_array, deg_list, win, bs_n, shared_list, nn, lock):
 			nn[1] += 1
 			print(str(nn[1]) + ' / ' + str(nn[0]))
 
+def thread_one_bs_pt(tfbs_array, deg_list, win, bs_n, shared_list, nn, lock):
+	tf_n = len(tfbs_array)
+	epsilon = 1e-15   # 常用值
+	for i in range(tf_n):
+		# fin_n += 1
+		gene_set_pt, gene_list_pt = permutation(tfbs_array[i][1], deg_list)
+		beta_xy, pval_xy, DTCI, bs_pval = rrho_coor_bs(gene_list_pt, gene_set_pt, win, bs_n)
+		rank_score = (DTCI * min(1.0, -np.log10(max(bs_pval, 1e-5)) / 5.0)) ** 0.5
+		with lock:
+			shared_list.append([tfbs_array[i][0], beta_xy, pval_xy, DTCI, bs_pval, rank_score])
+			nn[1] += 1
+			print(str(nn[1]) + ' / ' + str(nn[0]))
+
 def split_into_n_even(lst, n):
 	nn = len(lst)
 	base = nn // n		  # 每份基础大小
@@ -670,6 +713,62 @@ def olcr_bootstrap(set_names=None, list_file=None, win=30, thread_n=16, bs_n=200
 
 	return df
 
+def olcr_bootstrap_permutation(set_names=None, list_file=None, win=30, thread_n=16, bs_n=2000, data_dir=None):
+	if data_dir == None:
+		data_dir = overcast_data_dir
+	if not (os.path.exists(data_dir) and os.path.isdir(data_dir)):
+		print("OVERCAST_data folder can not found!")
+		exit(1)
+	sets_files = []
+	for set1 in set_names:
+		sets_file = data_dir + '/TF-target_sets/' + set1 + '.txt'
+		if not os.path.exists(sets_file):
+			print(sets_file, 'not exist!')
+			exit(1)
+		else:
+			sets_files.append(sets_file)
+	degl_file = list_file
+	if not os.path.exists(degl_file):
+		print(degl_file, 'not exist!')
+		exit(1)
+
+	tfbs_array = read_gene_sets(sets_files)
+	deg_list = read_DEG_list(degl_file)
+	
+	manager = mp.Manager()
+	shared_list = manager.list([])  # 支持任意类型
+	nn = manager.list([len(tfbs_array), 0])  # 支持任意类型
+	lock = manager.Lock()
+
+	tfbs_array_splits = split_into_n_even(tfbs_array, thread_n)
+	task_args = []
+	for ti in range(thread_n):
+		task_args.append((tfbs_array_splits[ti], deg_list, win, bs_n, shared_list, nn, lock))
+	
+	# 创建进程池
+	with mp.Pool(processes=thread_n) as pool:
+		print(f"thrend_n:", thread_n)
+		print(f"total {len(task_args)} tasks")
+		# starmap 会自动分配任务到进程池
+		start_time = time.time()
+		results = pool.starmap(thread_one_bs_pt, task_args)
+		elapsed = time.time() - start_time
+	# 打印结果
+	print("time consumed:", elapsed)
+
+	#sorted_results = sorted(shared_list, key=lambda x: (x[-1], -x[-2]), reverse=False) # lambda x: (x[4], -x[1]), reverse=False
+	sorted_results = sorted(shared_list, key=lambda x: x[-1], reverse=True) # lambda x: (x[4], -x[1]), reverse=False
+	results = []
+	ln = len(sorted_results)
+	for li in range(len(sorted_results)):
+		ratio = (li + 1) / ln
+		rline = sorted_results[li] + [ratio]
+		results.append(rline)
+	col_names = ['TF_motif', 'Beta3', 'Beta3_Pvalue', 'DTCI', 'DTCI_Pvalue', 'Rank_Score', 'Rank / N']
+	df = pd.DataFrame(results, columns=col_names)
+
+	return df
+
 def olcr_bootstrap_1cpu(set_names=None, list_file=None, win=30, thread_n=16, bs_n=2000, data_dir=None):
 	if data_dir == None:
 		data_dir = overcast_data_dir
@@ -866,6 +965,37 @@ def plot_contour(set_names=['human_hocomoco_CLink_wtcoor_1w_0.8'], list_file='in
 			plot_contour_plot(deg_list, tfbs_array[i][1], win)
 			break
 
+def plot_contour_permutation(set_names=['human_hocomoco_CLink_wtcoor_1w_0.8'], list_file='input_deg-list.txt', tf='NFKB1_HUMAN.H11MO.1.B', win=30, data_dir=None):
+	if data_dir == None:
+		data_dir = overcast_data_dir
+	if not (os.path.exists(data_dir) and os.path.isdir(data_dir)):
+		print("OVERCAST_data folder can not found!")
+		exit(1)
+	sets_files = []
+	for set1 in set_names:
+		sets_file = data_dir + '/TF-target_sets/' + set1 + '.txt'
+		if not os.path.exists(sets_file):
+			print(sets_file, 'not exist!')
+			exit(1)
+		else:
+			sets_files.append(sets_file)
+	degl_file = list_file
+	if not os.path.exists(degl_file):
+		print(degl_file, 'not exist!')
+		exit(1)
+
+	tfbs_array = read_gene_sets(sets_files)
+	deg_list = read_DEG_list(degl_file)
+
+	tf_n = len(tfbs_array)
+	epsilon = 1e-15   # 常用值
+	for i in range(tf_n):
+		if tfbs_array[i][0] == tf:
+			gene_set_pt, gene_list_pt = permutation(tfbs_array[i][1], deg_list)
+			plot_contour_plot(gene_list_pt, gene_set_pt, win)
+			# plot_contour_plot(deg_list, tfbs_array[i][1], win)
+			break
+
 if __name__ == '__main__':
 	'''
 	result = olcr(set_names=['human_jaspar_CLink_rci4_1w_0.8', 'human_hocomoco_CLink_rci4_1w_0.8'], list_file='OVERCAST_data/input_deg-list.txt', win=30, thread_n=16)

{pyovercast-1.0.8 → pyovercast-1.0.9}/PyOVERCAST.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: PyOVERCAST
-Version: 1.0.8
+Version: 1.0.9
 Summary: A Python package for mining key transcription factors from transcriptome data.
 Author-email: Tinghua Huang <thua45@126.com>
 License-Expression: MIT
@@ -78,6 +78,9 @@ if __name__ == '__main__':
 	# or predict one DEG-list with bootstrap
 	result = predict.olcr_bootstrap(set_names=['human_jaspar_CLink_rci4_1w_0.8', 'human_hocomoco_CLink_rci4_1w_0.8'], list_file='./OVERCAST_data/input_deg-list.txt', win=30, bs_n=1000, thread_n=32)
 
+	# or predict one DEG-list with bootstrap and permutation
+	result = predict.olcr_bootstrap_permutation(set_names=['human_jaspar_CLink_rci4_1w_0.8', 'human_hocomoco_CLink_rci4_1w_0.8'], list_file='./OVERCAST_data/input_deg-list.txt', win=30, bs_n=1000, thread_n=32)
+
 	# save result to text file
 	result.to_csv('output.txt', sep='\t', index=False, encoding='utf-8')
 	

{pyovercast-1.0.8 → pyovercast-1.0.9}/README.md

@@ -35,6 +35,9 @@ if __name__ == '__main__':
 	# or predict one DEG-list with bootstrap
 	result = predict.olcr_bootstrap(set_names=['human_jaspar_CLink_rci4_1w_0.8', 'human_hocomoco_CLink_rci4_1w_0.8'], list_file='./OVERCAST_data/input_deg-list.txt', win=30, bs_n=1000, thread_n=32)
 
+	# or predict one DEG-list with bootstrap and permutation
+	result = predict.olcr_bootstrap_permutation(set_names=['human_jaspar_CLink_rci4_1w_0.8', 'human_hocomoco_CLink_rci4_1w_0.8'], list_file='./OVERCAST_data/input_deg-list.txt', win=30, bs_n=1000, thread_n=32)
+
 	# save result to text file
 	result.to_csv('output.txt', sep='\t', index=False, encoding='utf-8')
 	

{pyovercast-1.0.8 → pyovercast-1.0.9}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "PyOVERCAST"
-version = "1.0.8"
+version = "1.0.9"
 license = "MIT"  # SPDX expression
 description = "A Python package for mining key transcription factors from transcriptome data."
 readme = "README.md"