PyTFBS 1.0.5__tar.gz → 1.0.7__tar.gz

{pytfbs-1.0.5 → pytfbs-1.0.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: PyTFBS
-Version: 1.0.5
+Version: 1.0.7
 Summary: PyTFBS: A Python Package for Transcription Factor Binding Site Prediction
 Author-email: Tinghua Huang <thua45@126.com>
 License-Expression: MIT
@@ -21,6 +21,12 @@ Requires-Dist: torch
 Requires-Dist: numpy==1.26; sys_platform == "darwin"
 Requires-Dist: numpy>=2.0; sys_platform == "win32"
 Requires-Dist: numpy>=2.0; sys_platform == "linux"
+Requires-Dist: pandas==3.0; sys_platform == "darwin"
+Requires-Dist: pandas>=3.0; sys_platform == "win32"
+Requires-Dist: pandas>=3.0; sys_platform == "linux"
+Requires-Dist: scipy==1.18; sys_platform == "darwin"
+Requires-Dist: scipy==1.18; sys_platform == "win32"
+Requires-Dist: scipy==1.18; sys_platform == "linux"
 Provides-Extra: dev
 Requires-Dist: matplotlib; extra == "dev"
 Provides-Extra: test
@@ -29,18 +35,20 @@ Dynamic: license-file
 
 # PyTFBS
 
-A Python package for dpredicting transcription factor binding sites.
+A Python package for predicting transcription factor binding sites.
 
 ## Installation
 
 ```bash
-pip install torch numpy PyTFBS
+pip install torch numpy pandas scipy PyTFBS
 ```
 
 ## Usage
 
 ```python
-from PyTFBS import motif, predict
+from PyTFBS import motif, predict, clink
+import random
+random.seed(42)
 
 # download PyTFBS data, only need to run once!!!
 motif.download_data()
@@ -66,4 +74,27 @@ predict.win_bin('CEBPB_HUMAN.H11MO.0.A', 'CEBPB_HUMAN.H11MO.0.A', 'input_seq_fil
 # the my_motif_dir should be organized as [[motif], [trace], [par]]
 predict.script('CEBPB_HUMAN.H11MO.0.A', 'CEBPB_HUMAN.H11MO.0.A', 'input_seq_file.fasta', data_dir='my_motif_dir')
 predict.win_bin('CEBPB_HUMAN.H11MO.0.A', 'CEBPB_HUMAN.H11MO.0.A', 'input_seq_file.fasta', 64, data_dir='my_motif_dir')
+
+# create the CLink TF-target set
+
+# rexp_files = get_rexp_files()
+# print(rexp_files)
+
+df_rexps = clink.read_rexp_file()
+tf_cnames = clink.read_tf_cname()
+tf_code = 'CREB1_HUMAN.H11MO.0.A_RC'
+tf_tars = clink.read_PyTFBS_output(tf_code + '_output.txt')
+
+for tf_cname in tf_cnames[tf_code]:
+	if tf_cname[1] not in df_rexps.index:
+		continue
+	else:
+		tf_exp = df_rexps.loc[tf_cname[1]].values.tolist()
+	df_rexps_tars = df_rexps[df_rexps.index.isin(tf_tars)]
+	tar_exps = {idx: row.tolist() for idx, row in df_rexps_tars.iterrows()}
+	rcis = clink.rci(tar_exps, tf_exp)
+	for rci in rcis:
+		score = rci[1] * (1 + (random.uniform(1E-5, 1.0) / 1E10))
+		print(tf_cname[0] + '\t' + rci[0] + '\t' + str(score))
+
 ```

pytfbs-1.0.7/PyTFBS/clink.py

@@ -0,0 +1,214 @@
+import os
+import math
+import pandas as pd
+import numpy as np
+np.random.seed(42)
+from collections import defaultdict
+import random
+random.seed(42)
+from scipy.stats import t
+
+PyTFBS_data_dir = './PyTFBS_data'
+
+def sign(x):
+	if x > 0:
+		return 1
+	elif x < 0:
+		return -1
+	else:
+		return 0
+
+def weighted_corr(x1, y1, alpha):
+	"""
+	计算加权皮尔森相关系数及其p值。
+	
+	权重由每个点的 (|x| + |y|) / 2 计算，并进一步按 alpha 次幂调整。
+	
+	参数:
+		x, y : array-like
+			输入变量。
+		alpha : float
+			权重调整指数，控制权重对极端值的敏感度。
+	
+	返回:
+		r : float
+			加权皮尔森相关系数。
+		p : float
+			双侧检验的p值，检验原假设为总体相关系数为0。
+			如果有效自由度 ≤ 0，则返回 p = np.nan。
+	"""
+	x = np.array(x1)
+	y = np.array(y1)
+	
+	# 计算基础权重（基于数值大小）
+	w = (np.abs(x) + np.abs(y)) / 2.0
+	# 调整权重，使其最大值缩放至1后再取alpha次幂
+	w_adj = (w / np.max(w)) ** alpha
+	# 归一化权重（和为1）
+	w_norm = w_adj / np.sum(w_adj)
+	
+	# 加权均值
+	x_bar = np.sum(w_norm * x)
+	y_bar = np.sum(w_norm * y)
+	
+	# 加权协方差与标准差
+	dx = x - x_bar
+	dy = y - y_bar
+	cov = np.sum(w_norm * dx * dy)
+	var_x = np.sum(w_norm * dx**2)
+	var_y = np.sum(w_norm * dy**2)
+	std_x = np.sqrt(var_x)
+	std_y = np.sqrt(var_y)
+	
+	if std_x * std_y == 0.0:
+		r = 0.0
+	else:
+		r = cov / (std_x * std_y)
+	
+	# ---- 计算p值（基于t分布近似） ----
+	# 有效样本量（权重之和的平方除以权重平方和）
+	sum_w = np.sum(w_adj)		  # 注意：w_adj未经归一化，但比例不变
+	sum_w2 = np.sum(w_adj**2)
+	n_eff = (sum_w ** 2) / sum_w2 if sum_w2 > 0 else 0.0
+	
+	df = n_eff - 2  # 自由度
+	if df > 0 and abs(r) < 1.0:
+		# t统计量：t = r * sqrt(df / (1 - r^2))
+		t_stat = r * np.sqrt(df / (1 - r**2))
+		p = 2 * (1 - t.cdf(abs(t_stat), df))
+	else:
+		p = np.nan  # 无法计算或相关系数为±1时，p值无意义
+	
+	return r, p
+
+def norm_std(data, sample=True):
+	"""
+	手动计算标准差
+	sample=True:  样本标准差（除以 n-1）
+	sample=False: 总体标准差（除以 n）
+	"""
+	n = len(data)
+	if n < 2:
+		raise ValueError("need 2 obsvs")
+	
+	# 步骤 1：计算平均值
+	# mean = sum(data) / n
+	
+	# 步骤 2 & 3：计算每个数据与平均值的差的平方
+	# squared_diffs = [(x - mean) ** 2 for x in data]
+	squared_diffs = [x ** 2 for x in data]
+	
+	# 步骤 4：计算方差
+	if sample:
+		variance = sum(squared_diffs) / (n - 1)  # 样本方差
+	else:
+		variance = sum(squared_diffs) / n  # 总体方差
+	
+	# 步骤 5：开平方根得到标准差
+	std = variance ** 0.5
+	
+	k = max(1, int(len(data) * 0.05))  # 至少取1个
+	sorted_lst = sorted([abs(i) for i in data], reverse=True)
+	max_obsv = sorted_lst[k - 1]
+	std_norm = std / max_obsv
+
+	return std_norm
+
+def rci(tar_exps, tf_exp):
+	tar_rcis = []
+	for tar_name in tar_exps.keys():
+		tar_exp = tar_exps[tar_name]
+		corr_wt, p_value = weighted_corr(tar_exp, tf_exp, alpha=1.0)
+		tar_cv = norm_std(tar_exp)
+		if math.isnan(tar_cv):
+			tar_cv = 0.0
+		tf_cv = norm_std(tf_exp)
+		if math.isnan(tf_cv):
+			tf_cv = 0.0
+		if math.isnan(corr_wt) or p_value >= 0.05:
+			continue
+		else:
+			rci = sign(corr_wt) * (abs(corr_wt)**1.0 * tar_cv**0.5 * tf_cv**0.5) ** (1.0 / (1.0 + 0.5 + 0.5))
+			tar_rcis.append((tar_name, rci))
+	return tar_rcis
+
+def read_tf_cname(file=None, data_dir=None):
+	if data_dir == None:
+		data_dir = PyTFBS_data_dir
+	if file == None:
+		file = data_dir + '/TF-code-name.txt'
+	else:
+		file = data_dir + '/' + file
+	if not os.path.exists(file):
+		print('TF-code-name.txt not exist, you may need to run download_data() first')
+		exit(1)
+	tf_cnames = defaultdict(list)
+	for line in open(file, 'r'):
+		lblocks = line.rstrip().split('\t')
+		tf_cnames[lblocks[1]].append((lblocks[2], lblocks[3]))
+	return tf_cnames
+
+def read_rexp_file(file=None, data_dir=None):
+	if data_dir == None:
+		data_dir = PyTFBS_data_dir + '/Ref_Exps'
+	if file == None:
+		file = data_dir + '/Human_ED_matrix_1.5.txt'
+	else:
+		file = data_dir + '/' + file_name
+	if not os.path.exists(file):
+		print('Human_ED_matrix_1.5.txt not exist, you may need to run download_data() first')
+		exit(1)
+	df_rexps = pd.read_csv(file, sep='\t', header=0, index_col=0)
+	return df_rexps
+
+def read_PyTFBS_output(file):
+	tf_tars = set()
+	for line in open(file):
+		if line[0] == '#':
+			continue
+		lblocks = line.rstrip().split('\t')
+		gene_name = lblocks[0].split('::')[0].split('|')[2]
+		score = float(lblocks[3])
+		jindex = float(lblocks[2])
+		if score > 0.8 and jindex > 0.3:
+			tf_tars.add(gene_name)
+	return tf_tars
+
+def get_rexp_files(species=None, data_dir=None):
+	if data_dir == None:
+		data_dir = PyTFBS_data_dir
+	index_file = data_dir + '/RExps_index.txt'
+	if not os.path.exists(index_file):
+		print('RExps_index.txt not exist, you may need to run download_data() first')
+		exit(1)
+	header = ''
+	result = []
+	for line in open(index_file, 'r'):
+		if line[0] == "#":
+			header = line.rstrip()
+			continue
+		lblocks = line.rstrip().split('\t')
+		if species != None and lblocks[1] != species:
+			continue
+		result.append(lblocks[0])
+	return result
+
+if __name__ == '__main__':
+	rexp_files = get_rexp_files()
+	print(rexp_files)
+	df_rexps = read_rexp_file()
+	tf_cnames = read_tf_cname()
+	tf_code = 'CREB1_HUMAN.H11MO.0.A_RC'
+	tf_tars = read_PyTFBS_output(tf_code + '_output.txt')
+
+	for tf_cname in tf_cnames[tf_code]:
+		if tf_cname[1] not in df_rexps.index:
+			continue
+		else:
+			tf_exp = df_rexps.loc[tf_cname[1]].values.tolist()
+		df_rexps_tars = df_rexps[df_rexps.index.isin(tf_tars)]
+		tar_exps = {idx: row.tolist() for idx, row in df_rexps_tars.iterrows()}
+		rcis = rci(tar_exps, tf_exp)
+		for rci in rcis:
+			score = rci[1] * (1 + (random.uniform(1E-5, 1.0) / 1E10))
+			print(tf_cname[0] + '\t' + rci[0] + '\t' + str(score))

{pytfbs-1.0.5 → pytfbs-1.0.7}/PyTFBS.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: PyTFBS
-Version: 1.0.5
+Version: 1.0.7
 Summary: PyTFBS: A Python Package for Transcription Factor Binding Site Prediction
 Author-email: Tinghua Huang <thua45@126.com>
 License-Expression: MIT
@@ -21,6 +21,12 @@ Requires-Dist: torch
 Requires-Dist: numpy==1.26; sys_platform == "darwin"
 Requires-Dist: numpy>=2.0; sys_platform == "win32"
 Requires-Dist: numpy>=2.0; sys_platform == "linux"
+Requires-Dist: pandas==3.0; sys_platform == "darwin"
+Requires-Dist: pandas>=3.0; sys_platform == "win32"
+Requires-Dist: pandas>=3.0; sys_platform == "linux"
+Requires-Dist: scipy==1.18; sys_platform == "darwin"
+Requires-Dist: scipy==1.18; sys_platform == "win32"
+Requires-Dist: scipy==1.18; sys_platform == "linux"
 Provides-Extra: dev
 Requires-Dist: matplotlib; extra == "dev"
 Provides-Extra: test
@@ -29,18 +35,20 @@ Dynamic: license-file
 
 # PyTFBS
 
-A Python package for dpredicting transcription factor binding sites.
+A Python package for predicting transcription factor binding sites.
 
 ## Installation
 
 ```bash
-pip install torch numpy PyTFBS
+pip install torch numpy pandas scipy PyTFBS
 ```
 
 ## Usage
 
 ```python
-from PyTFBS import motif, predict
+from PyTFBS import motif, predict, clink
+import random
+random.seed(42)
 
 # download PyTFBS data, only need to run once!!!
 motif.download_data()
@@ -66,4 +74,27 @@ predict.win_bin('CEBPB_HUMAN.H11MO.0.A', 'CEBPB_HUMAN.H11MO.0.A', 'input_seq_fil
 # the my_motif_dir should be organized as [[motif], [trace], [par]]
 predict.script('CEBPB_HUMAN.H11MO.0.A', 'CEBPB_HUMAN.H11MO.0.A', 'input_seq_file.fasta', data_dir='my_motif_dir')
 predict.win_bin('CEBPB_HUMAN.H11MO.0.A', 'CEBPB_HUMAN.H11MO.0.A', 'input_seq_file.fasta', 64, data_dir='my_motif_dir')
+
+# create the CLink TF-target set
+
+# rexp_files = get_rexp_files()
+# print(rexp_files)
+
+df_rexps = clink.read_rexp_file()
+tf_cnames = clink.read_tf_cname()
+tf_code = 'CREB1_HUMAN.H11MO.0.A_RC'
+tf_tars = clink.read_PyTFBS_output(tf_code + '_output.txt')
+
+for tf_cname in tf_cnames[tf_code]:
+	if tf_cname[1] not in df_rexps.index:
+		continue
+	else:
+		tf_exp = df_rexps.loc[tf_cname[1]].values.tolist()
+	df_rexps_tars = df_rexps[df_rexps.index.isin(tf_tars)]
+	tar_exps = {idx: row.tolist() for idx, row in df_rexps_tars.iterrows()}
+	rcis = clink.rci(tar_exps, tf_exp)
+	for rci in rcis:
+		score = rci[1] * (1 + (random.uniform(1E-5, 1.0) / 1E10))
+		print(tf_cname[0] + '\t' + rci[0] + '\t' + str(score))
+
 ```

{pytfbs-1.0.5 → pytfbs-1.0.7}/PyTFBS.egg-info/SOURCES.txt

@@ -2,6 +2,7 @@ LICENSE
 README.md
 pyproject.toml
 PyTFBS/__init__.py
+PyTFBS/clink.py
 PyTFBS/motif.py
 PyTFBS/predict.py
 PyTFBS.egg-info/PKG-INFO

{pytfbs-1.0.5 → pytfbs-1.0.7}/PyTFBS.egg-info/requires.txt

@@ -2,12 +2,18 @@ torch
 
 [:sys_platform == "darwin"]
 numpy==1.26
+pandas==3.0
+scipy==1.18
 
 [:sys_platform == "linux"]
 numpy>=2.0
+pandas>=3.0
+scipy==1.18
 
 [:sys_platform == "win32"]
 numpy>=2.0
+pandas>=3.0
+scipy==1.18
 
 [dev]
 matplotlib

{pytfbs-1.0.5 → pytfbs-1.0.7}/README.md

@@ -1,17 +1,19 @@
 # PyTFBS
 
-A Python package for dpredicting transcription factor binding sites.
+A Python package for predicting transcription factor binding sites.
 
 ## Installation
 
 ```bash
-pip install torch numpy PyTFBS
+pip install torch numpy pandas scipy PyTFBS
 ```
 
 ## Usage
 
 ```python
-from PyTFBS import motif, predict
+from PyTFBS import motif, predict, clink
+import random
+random.seed(42)
 
 # download PyTFBS data, only need to run once!!!
 motif.download_data()
@@ -37,4 +39,27 @@ predict.win_bin('CEBPB_HUMAN.H11MO.0.A', 'CEBPB_HUMAN.H11MO.0.A', 'input_seq_fil
 # the my_motif_dir should be organized as [[motif], [trace], [par]]
 predict.script('CEBPB_HUMAN.H11MO.0.A', 'CEBPB_HUMAN.H11MO.0.A', 'input_seq_file.fasta', data_dir='my_motif_dir')
 predict.win_bin('CEBPB_HUMAN.H11MO.0.A', 'CEBPB_HUMAN.H11MO.0.A', 'input_seq_file.fasta', 64, data_dir='my_motif_dir')
+
+# create the CLink TF-target set
+
+# rexp_files = get_rexp_files()
+# print(rexp_files)
+
+df_rexps = clink.read_rexp_file()
+tf_cnames = clink.read_tf_cname()
+tf_code = 'CREB1_HUMAN.H11MO.0.A_RC'
+tf_tars = clink.read_PyTFBS_output(tf_code + '_output.txt')
+
+for tf_cname in tf_cnames[tf_code]:
+	if tf_cname[1] not in df_rexps.index:
+		continue
+	else:
+		tf_exp = df_rexps.loc[tf_cname[1]].values.tolist()
+	df_rexps_tars = df_rexps[df_rexps.index.isin(tf_tars)]
+	tar_exps = {idx: row.tolist() for idx, row in df_rexps_tars.iterrows()}
+	rcis = clink.rci(tar_exps, tf_exp)
+	for rci in rcis:
+		score = rci[1] * (1 + (random.uniform(1E-5, 1.0) / 1E10))
+		print(tf_cname[0] + '\t' + rci[0] + '\t' + str(score))
+
 ```

{pytfbs-1.0.5 → pytfbs-1.0.7}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "PyTFBS"
-version = "1.0.5"
+version = "1.0.7"
 license = "MIT"  # SPDX expression
 description = "PyTFBS: A Python Package for Transcription Factor Binding Site Prediction"
 readme = "README.md"
@@ -25,6 +25,12 @@ dependencies = [
     "numpy==1.26; sys_platform == 'darwin'",
     "numpy>=2.0; sys_platform == 'win32'",
     "numpy>=2.0; sys_platform == 'linux'",
+    "pandas==3.0; sys_platform == 'darwin'",
+    "pandas>=3.0; sys_platform == 'win32'",
+    "pandas>=3.0; sys_platform == 'linux'",
+    "scipy==1.18; sys_platform == 'darwin'",
+    "scipy==1.18; sys_platform == 'win32'",
+    "scipy==1.18; sys_platform == 'linux'",
 ]
 
 [project.urls]