import pandas as pd
import numpy as np
from tqdm import tqdm
import helper
pathways, interactome = helper.setup()
interactome_e2n = helper.convert_edges_to_node(interactome, 'edge_weight',
'interactome_weight')
interactome_en = helper.keep_edge_nodes(interactome_e2n,
['head', 'interactome_weight'])
interactome_degrees = pd.read_csv(
'../output/features_interactome_no_nearest_01.txt', delimiter='\t')
interactome_features = pd.merge(interactome_degrees,
interactome_en,
left_on='name',
right_on='head')
interactome_features.drop('head', axis=1, inplace=True)
num_folds = 2
create_additional_featuers = False
for pathway in tqdm(pathways):
pathway_dist_score = pd.read_csv(
'../output/features_{}_03.txt'.format(pathway),
delimiter='\t',
"TCR",
"TGF_beta_Receptor",
"TNFalpha",
"Wnt"]
interactome = pd.read_csv(
'../data/pathlinker-signaling-children-reg-weighted.txt',
delimiter='\t')
return pathways, interactome
#
# Begin Script
#
pathways, interactome = setup()
interactome_e2n = helper.convert_edges_to_node(interactome,
'edge_weight',
'interactome_weight')
# pathlinker_weight_attributes = {}
# pagerank_weight_attributes = {}
# interactome_weight_attributes = {}
for pathway_name in tqdm.tqdm(pathways):
# pathway_df = pd.read_csv('../data/pathways/{}-edges.txt'.
# format(pathway_name),
# delimiter='\t')
# pathway_df_n2e = helper.convert_edges_to_node(pathway_df)
# node_features_df = '../output/features_{}'.format(pathway_name)
pathways, interactome = setup()
for pathway in tqdm(pathways):
pathway_dist_en = pd.read_csv('../output/features_{}_02.txt'.
format(pathway),
delimiter='\t')
pagerank_df = pd.read_csv('../data/pagerank/{}-q_0.50-edge-fluxes.txt'.
format(pathway),
delimiter='\t')
cyclinker_df = pd.read_csv('../data/cyclinker/{}-k_110000-ranked-edges.txt'.
format(pathway),
delimiter='\t')
pagerank_df_e2n = helper.convert_edges_to_node(
pagerank_df, 'edge_flux', 'pagerank_value')
cyclinker_df_e2n = helper.convert_edges_to_node(
cyclinker_df, 'KSP index', 'cyclinker_value')
pagerank_en = helper.keep_edge_nodes(
pagerank_df_e2n, ['head', 'pagerank_value'])
cyclinker_en = helper.keep_edge_nodes(
cyclinker_df_e2n, ['head', 'cyclinker_value'])
pathway_dist_ranks = pd.merge(
pathway_dist_en, pagerank_en,
left_on='name', right_on='head', how='left')
pathway_dist_ranks = pd.merge(
pathway_dist_ranks, cyclinker_en,
left_on='name', right_on='head', how='left')
def main():
pathways, interactome = setup()
# I f****d up.
# this file should not loop over any of the pathways
# the features here should be calculated on the entire interactome
# this is why the variable names will be confusing
# for pathway_name in pathways:
# pathway_df = pd.read_csv('../data/pathways/{}-edges.txt'.
# format(pathway_name),
# delimiter='\t')
pathway_name = 'interactome'
pathway_df = interactome
pathway_df_n2e = helper.convert_edges_to_node(pathway_df,
weight_col='edge_weight')
g = nx.from_pandas_dataframe(pathway_df_n2e,
source='#tail',
target='head',
create_using=nx.DiGraph())
# print('calculating betweenness_attributes')
# betweenness_attributes = cna.calculate_betweenesss(g)
# print('finished betweenness_attributes')
print('calculating degree_attributes')
degree_attributes = cna.calculate_degree(g)
print('finished degree_attributes')
# print('calculating katz_attributes')
# katz_attributes = cna.calculate_katz(g)
# print('finished katz_attributes')
# nx.set_node_attributes(g, 'betweenness', betweenness_attributes)
nx.set_node_attributes(g, 'degree', degree_attributes)
# nx.set_node_attributes(g, 'katz', katz_attributes)
# nearest_1_attributes = {} # c
# nearest_3_attributes = {} # c
# nearest_5_attributes = {}
max_degree_head_tail_attributes = {} # c
min_degree_head_tail_attributes = {} # c
avg_degree_head_tail_attributes = {} # c
edge_node_pattern = re.compile('.*_to_.*')
print('iterating over nodes')
for node in tqdm.tqdm(g.nodes_iter()):
match = edge_node_pattern.match(node)
if match:
# n1 = cna.calculate_nearest_k_nodes(g, node, 2)
# nearest_1_attributes[node] = n1
# n3 = cna.calculate_nearest_k_nodes(g, node, 4)
# nearest_3_attributes[node] = n3
# n5 = cna.calculate_nearest_k_nodes(g, node, 5)
# nearest_5_attributes[node] = n5
max_d_ht = cna.calculate_max_degree_head_tail(g, node)
max_degree_head_tail_attributes[node] = max_d_ht
min_d_ht = cna.calculate_min_degree_head_tail(g, node)
min_degree_head_tail_attributes[node] = min_d_ht
avg_d_ht = cna.calculate_avg_degree_head_tail(g, node)
avg_degree_head_tail_attributes[node] = avg_d_ht
print('setting node attributes')
# nx.set_node_attributes(g, 'nearest_1', nearest_1_attributes)
# nx.set_node_attributes(g, 'nearest_3', nearest_3_attributes)
# nx.set_node_attributes(g, 'nearest_5', nearest_5_attributes)
nx.set_node_attributes(g, 'max_degree_head_tail',
max_degree_head_tail_attributes)
nx.set_node_attributes(g, 'min_degree_head_tail',
min_degree_head_tail_attributes)
nx.set_node_attributes(g, 'avg_degree_head_tail',
avg_degree_head_tail_attributes)
node_filename = '../output/features_{}_no_nearest_01.txt'.format(
pathway_name)
with open(node_filename, 'w') as f:
col_names = [
'name',
# 'betweenness',
'degree',
# 'katz',
# 'nearest_1',
# 'nearest_3',
# 'nearest_5',
'max_degree_head_tail',
'min_degree_head_tail',
'avg_degree_head_tail'
]
fstring = ['%s'] * len(col_names)
fstring = '\t'.join(fstring) + '\n'
wstring = fstring % tuple(col_names)
f.write(wstring)
print('writing node info')
for node in tqdm.tqdm(g.nodes_iter()):
match = edge_node_pattern.match(node)
if match:
na = g.node[node]
f.write(fstring % tuple([
node,
# na['betweenness'],
na['degree'],
# na['katz'],
# na['nearest_1'],
# na['nearest_3'],
# na['nearest_5'],
na['max_degree_head_tail'],
na['min_degree_head_tail'],
na['avg_degree_head_tail']
]))
print("{} created".format(node_filename))
def main():
pathways, interactome = setup()
# I f****d up.
# this file should not loop over any of the pathways
# the features here should be calculated on the entire interactome
# this is why the variable names will be confusing
# for pathway_name in pathways:
# pathway_df = pd.read_csv('../data/pathways/{}-edges.txt'.
# format(pathway_name),
# delimiter='\t')
pathway_name = 'interactome'
pathway_df = interactome
pathway_df_n2e = helper.convert_edges_to_node(
pathway_df, weight_col='edge_weight')
g = nx.from_pandas_dataframe(
pathway_df_n2e, source='#tail', target='head',
create_using=nx.DiGraph())
# print('calculating betweenness_attributes')
# betweenness_attributes = cna.calculate_betweenesss(g)
# print('finished betweenness_attributes')
print('calculating degree_attributes')
degree_attributes = cna.calculate_degree(g)
print('finished degree_attributes')
# print('calculating katz_attributes')
# katz_attributes = cna.calculate_katz(g)
# print('finished katz_attributes')
# nx.set_node_attributes(g, 'betweenness', betweenness_attributes)
nx.set_node_attributes(g, 'degree', degree_attributes)
# nx.set_node_attributes(g, 'katz', katz_attributes)
# nearest_1_attributes = {} # c
# nearest_3_attributes = {} # c
# nearest_5_attributes = {}
max_degree_head_tail_attributes = {} # c
min_degree_head_tail_attributes = {} # c
avg_degree_head_tail_attributes = {} # c
edge_node_pattern = re.compile('.*_to_.*')
print('iterating over nodes')
for node in tqdm.tqdm(g.nodes_iter()):
match = edge_node_pattern.match(node)
if match:
# n1 = cna.calculate_nearest_k_nodes(g, node, 2)
# nearest_1_attributes[node] = n1
# n3 = cna.calculate_nearest_k_nodes(g, node, 4)
# nearest_3_attributes[node] = n3
# n5 = cna.calculate_nearest_k_nodes(g, node, 5)
# nearest_5_attributes[node] = n5
max_d_ht = cna.calculate_max_degree_head_tail(g, node)
max_degree_head_tail_attributes[node] = max_d_ht
min_d_ht = cna.calculate_min_degree_head_tail(g, node)
min_degree_head_tail_attributes[node] = min_d_ht
avg_d_ht = cna.calculate_avg_degree_head_tail(g, node)
avg_degree_head_tail_attributes[node] = avg_d_ht
print('setting node attributes')
# nx.set_node_attributes(g, 'nearest_1', nearest_1_attributes)
# nx.set_node_attributes(g, 'nearest_3', nearest_3_attributes)
# nx.set_node_attributes(g, 'nearest_5', nearest_5_attributes)
nx.set_node_attributes(g, 'max_degree_head_tail',
max_degree_head_tail_attributes)
nx.set_node_attributes(g, 'min_degree_head_tail',
min_degree_head_tail_attributes)
nx.set_node_attributes(g, 'avg_degree_head_tail',
avg_degree_head_tail_attributes)
node_filename = '../output/features_{}_no_nearest_01.txt'.format(
pathway_name)
with open(node_filename, 'w') as f:
col_names = ['name',
# 'betweenness',
'degree',
# 'katz',
# 'nearest_1',
# 'nearest_3',
# 'nearest_5',
'max_degree_head_tail', 'min_degree_head_tail',
'avg_degree_head_tail']
fstring = ['%s'] * len(col_names)
fstring = '\t'.join(fstring) + '\n'
wstring = fstring % tuple(col_names)
f.write(wstring)
print('writing node info')
for node in tqdm.tqdm(g.nodes_iter()):
match = edge_node_pattern.match(node)
if match:
na = g.node[node]
f.write(fstring % tuple([
node,
# na['betweenness'],
na['degree'],
# na['katz'],
# na['nearest_1'],
# na['nearest_3'],
# na['nearest_5'],
na['max_degree_head_tail'], na['min_degree_head_tail'],
na['avg_degree_head_tail']
]))
print("{} created".format(node_filename))