annotate build_protein_interaction_maps.py @ 0:3fcdc585cd2e draft

planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
author proteore
date Fri, 01 Mar 2019 11:23:22 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
1 # -*- coding: utf-8 -*-
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
2 import csv, json, argparse, re
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
3
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
4 def get_args() :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
5 parser = argparse.ArgumentParser()
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
6 parser.add_argument("--species")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
7 parser.add_argument("--database", help="Humap, Bioplex or Biogrid", required=True)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
8 parser.add_argument("--dict_path", required=True)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
9 parser.add_argument("--input_type", help="type of input (list of id or filename)",required=True)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
10 parser.add_argument("--input", required=True)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
11 parser.add_argument("--header")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
12 parser.add_argument("--ncol")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
13 parser.add_argument("--id_type")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
14 parser.add_argument("--network_output")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
15 parser.add_argument("--nodes_output")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
16 args = parser.parse_args()
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
17
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
18 if args.input_type=="file" :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
19 args.ncol = nb_col_to_int(args.ncol)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
20 args.header = str2bool(args.header)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
21
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
22 return args
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
23
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
24 #Turn string into boolean
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
25 def str2bool(v):
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
26 if v.lower() in ('yes', 'true', 't', 'y', '1'):
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
27 return True
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
28 elif v.lower() in ('no', 'false', 'f', 'n', '0'):
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
29 return False
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
30 else:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
31 raise argparse.ArgumentTypeError('Boolean value expected.')
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
32
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
33 #return the column number in int format
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
34 def nb_col_to_int(nb_col):
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
35 try :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
36 nb_col = int(nb_col.replace("c", "")) - 1
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
37 return nb_col
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
38 except :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
39 sys.exit("Please specify the column where you would like to apply the filter with valid format")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
40
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
41 #return list of (unique) ids from string
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
42 def get_input_ids_from_string(input) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
43 ids_list = list(set(re.split(r'\s+',input.replace("\r","").replace("\n"," ").replace("\t"," "))))
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
44 if "" in ids_list : ids_list.remove("")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
45 #if "NA" in ids_list : ids_list.remove("NA")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
46 return ids_list
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
47
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
48 #return input_file and list of unique ids from input file path
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
49 def get_input_ids_from_file(input,nb_col,header) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
50 with open(input, "r") as csv_file :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
51 input_file= list(csv.reader(csv_file, delimiter='\t'))
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
52
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
53 input_file, ids_list = one_id_one_line(input_file,nb_col,header)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
54 if "" in ids_list : ids_list.remove("")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
55 #if "NA" in ids_list : ids_list.remove("NA")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
56
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
57 return input_file, ids_list
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
58
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
59 #return input file by adding lines when there are more than one id per line
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
60 def one_id_one_line(input_file,nb_col,header) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
61
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
62 if header :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
63 new_file = [input_file[0]]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
64 input_file = input_file[1:]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
65 else :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
66 new_file=[]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
67 ids_list=[]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
68
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
69 for line in input_file :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
70 if line != [] and set(line) != {''}:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
71 line[nb_col] = re.sub(r"\s+","",line[nb_col])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
72 if ";" in line[nb_col] :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
73 ids = line[nb_col].split(";")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
74 for id in ids :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
75 new_file.append(line[:nb_col]+[id]+line[nb_col+1:])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
76 ids_list.append(id)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
77 else :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
78 new_file.append(line)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
79 ids_list.append(line[nb_col])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
80
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
81 ids_list= list(set(ids_list))
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
82
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
83 return new_file, ids_list
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
84
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
85 #replace all blank cells to NA
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
86 def blank_to_NA(csv_file) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
87 tmp=[]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
88 for line in csv_file :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
89 line = ["NA" if cell=="" or cell==" " or cell=="NaN" or cell=="-" else cell for cell in line]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
90 tmp.append(line)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
91
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
92 return tmp
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
93
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
94 def biogrid_output_files(ids,species) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
95 network_file=[["Entrez Gene Interactor A","Entrez Gene Interactor B","Gene symbol Interactor A","Gene symbol Interactor B","Experimental System","Experimental Type","Pubmed ID","Interaction Score","Phenotypes"]]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
96 ids_set= set(ids)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
97 ids_not_found=set([])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
98 for id in ids :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
99 if id in ppi_dict['network'] :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
100 network_file.extend(ppi_dict['network'][id])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
101 ids_set.update([interact[1] for interact in ppi_dict['network'][id]])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
102 else :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
103 ids_not_found.add(id)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
104
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
105 nodes_file = [["Entrez gene ID","Official Symbol Interactor","Present in user input ids","ID present in Biogrid "+species,"Pathway"]]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
106 for id in ids_set:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
107 #get pathway
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
108 if id in ppi_dict['nodes']:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
109 description_pathway=";".join(ppi_dict['nodes'][id])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
110 else :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
111 description_pathway="NA"
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
112
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
113 #get gene name
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
114 if id in ppi_dict['network']: gene_name = ppi_dict['network'][id][0][2]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
115 else : gene_name="NA"
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
116
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
117 #make line
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
118 nodes_file.append([id]+[gene_name]+[id in ids]+[id not in ids_not_found]+[description_pathway])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
119
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
120 return network_file,nodes_file
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
121
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
122 def bioplex_output_files(ids,id_type,species) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
123 network_file=[[id_type+" Interactor A",id_type+" Interactor B","Gene symbol Interactor A","Gene symbol Interactor B","Interaction Score"]]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
124 ids_set= set(ids)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
125 ids_not_found=set([])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
126 for id in ids :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
127 if id in ppi_dict['network'][id_type] :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
128 network_file.extend(ppi_dict['network'][id_type][id])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
129 ids_set.update([interact[1] for interact in ppi_dict['network'][id_type][id]])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
130 else :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
131 ids_not_found.add(id)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
132
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
133 if id_type=="UniProt-AC" : nodes_file=[[id_type,"Present in user input ids","ID present in Human Bioplex","Pathway"]]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
134 else: nodes_file=[[id_type,"Official symbol Interactor","Present in user input ids","Present in interactome","Pathway"]]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
135 for id in ids_set:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
136
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
137 if id in ppi_dict['nodes'][id_type]:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
138 description_pathway=";".join(ppi_dict['nodes'][id_type][id])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
139 else :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
140 description_pathway="NA"
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
141
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
142 #make line
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
143 if id_type=="UniProt-AC":
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
144 nodes_file.append([id]+[id in ids]+[id not in ids_not_found]+[description_pathway])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
145 elif id_type=="GeneID":
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
146 #get gene_name
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
147 if id in ppi_dict['network'][id_type]: gene_name = ppi_dict['network'][id_type][id][0][2]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
148 else : gene_name="NA"
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
149 nodes_file.append([id]+[gene_name]+[id in ids]+[id not in ids_not_found]+[description_pathway])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
150
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
151 return network_file,nodes_file
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
152
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
153 def humap_output_files(ids,species) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
154 network_file=[["Entrez Gene Interactor A","Entrez Gene Interactor B","Gene symbol Interactor A","Gene symbol Interactor B","Interaction Score"]]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
155 ids_set= set(ids)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
156 ids_not_found=set([])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
157 for id in ids :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
158 if id in ppi_dict['network'] :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
159 network_file.extend(ppi_dict['network'][id])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
160 ids_set.update([interact[1] for interact in ppi_dict['network'][id]])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
161 else :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
162 ids_not_found.add(id)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
163
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
164 nodes_file = [["Entrez gene ID","Official Symbol Interactor","Present in user input ids","ID present in Hu.MAP","Pathway"]]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
165 for id in ids_set:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
166 if id in ppi_dict['nodes']:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
167 description_pathway=";".join(ppi_dict['nodes'][id])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
168 else :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
169 description_pathway="NA"
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
170
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
171 #get gene name
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
172 if id in ppi_dict['gene_name']:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
173 gene_name = ppi_dict['gene_name'][id]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
174 else :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
175 gene_name = "NA"
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
176
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
177 #make line
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
178 nodes_file.append([id]+[gene_name]+[id in ids]+[id not in ids_not_found]+[description_pathway])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
179
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
180 return network_file,nodes_file
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
181
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
182 #function to sort the csv_file by value in a specific column
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
183 def sort_by_column(tab,sort_col,reverse,header):
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
184
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
185 if len(tab) > 1 : #if there's more than just a header or 1 row
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
186 if header :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
187 head=tab[0]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
188 tab=tab[1:]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
189
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
190 #list of empty cells in the column to sort
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
191 unsortable_lines = [i for i,line in enumerate(tab) if (line[sort_col]=='' or line[sort_col] == 'NA')]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
192 unsorted_tab=[ tab[i] for i in unsortable_lines]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
193 tab= [line for i,line in enumerate(tab) if i not in unsortable_lines]
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
194
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
195 if only_number(tab,sort_col) and any_float(tab,sort_col) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
196 tab = sorted(tab, key=lambda row: float(row[sort_col]), reverse=reverse)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
197 elif only_number(tab,sort_col):
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
198 tab = sorted(tab, key=lambda row: int(row[sort_col]), reverse=reverse)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
199 else :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
200 tab = sorted(tab, key=lambda row: row[sort_col], reverse=reverse)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
201
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
202 tab.extend(unsorted_tab)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
203 if header is True : tab = [head]+tab
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
204
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
205 return tab
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
206
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
207 def only_number(tab,col) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
208
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
209 for line in tab :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
210 if not (is_number("float",line[col].replace(",",".")) or is_number("int",line[col].replace(",","."))) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
211 return False
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
212 return True
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
213
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
214 #Check if a variable is a float or an integer
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
215 def is_number(number_format, n):
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
216 float_format = re.compile(r"^[-]?[0-9][0-9]*.?[0-9]+$")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
217 int_format = re.compile(r"^[-]?[0-9][0-9]*$")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
218 test = ""
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
219 if number_format == "int":
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
220 test = re.match(int_format, n)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
221 elif number_format == "float":
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
222 test = re.match(float_format, n)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
223 if test:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
224 return True
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
225
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
226 #return True is there is at least one float in the column
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
227 def any_float(tab,col) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
228
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
229 for line in tab :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
230 if is_number("float",line[col].replace(",",".")) :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
231 return True
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
232
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
233 return False
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
234
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
235 def main() :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
236
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
237 #Get args from command line
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
238 global args
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
239 args = get_args()
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
240
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
241 #get PPI dictionary
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
242 with open(args.dict_path, 'r') as handle:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
243 global ppi_dict
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
244 ppi_dict = json.load(handle)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
245
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
246 #Get file and/or ids from input
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
247 if args.input_type == "text" :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
248 ids = get_input_ids_from_string(args.input)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
249 elif args.input_type == "file" :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
250 input_file, ids = get_input_ids_from_file(args.input,args.ncol,args.header)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
251
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
252 #create output files
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
253 if args.database=="biogrid":
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
254 network_file, nodes_file = biogrid_output_files(ids,args.species)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
255 elif args.database=="bioplex":
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
256 network_file, nodes_file = bioplex_output_files(ids,args.id_type,args.species)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
257 elif args.database=="humap":
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
258 network_file, nodes_file = humap_output_files(ids,args.species)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
259
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
260 #convert blank to NA and sort files
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
261 network_file = blank_to_NA(network_file)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
262 network_file = sort_by_column(network_file,0,False,True)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
263 nodes_file = sort_by_column(nodes_file,0,False,True)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
264
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
265 #write output files
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
266 with open(args.network_output,"w") as output :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
267 writer = csv.writer(output,delimiter="\t")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
268 writer.writerows(network_file)
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
269
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
270 with open(args.nodes_output,"w") as output :
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
271 writer = csv.writer(output,delimiter="\t")
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
272 for row in nodes_file:
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
273 writer.writerow([unicode(s).encode("utf-8") for s in row])
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
274
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
275 if __name__ == "__main__":
3fcdc585cd2e planemo upload commit 696e05eadff826a980b1150be5db5d1f52d06b1b-dirty
proteore
parents:
diff changeset
276 main()