Mercurial > repos > proteore > proteore_build_protein_interaction_maps
annotate build_protein_interaction_maps.py @ 0:b0ac71686b99 draft
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
author | proteore |
---|---|
date | Tue, 12 Mar 2019 05:55:54 -0400 |
parents | |
children | 0a85d709c4ae |
rev | line source |
---|---|
0
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
1 # -*- coding: utf-8 -*- |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
2 import csv, json, argparse, re |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
3 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
4 def get_args() : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
5 parser = argparse.ArgumentParser() |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
6 parser.add_argument("--species") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
7 parser.add_argument("--database", help="Humap, Bioplex or Biogrid", required=True) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
8 parser.add_argument("--dict_path", required=True) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
9 parser.add_argument("--input_type", help="type of input (list of id or filename)",required=True) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
10 parser.add_argument("--input", required=True) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
11 parser.add_argument("--header") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
12 parser.add_argument("--ncol") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
13 parser.add_argument("--id_type") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
14 parser.add_argument("--network_output") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
15 parser.add_argument("--nodes_output") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
16 args = parser.parse_args() |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
17 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
18 if args.input_type=="file" : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
19 args.ncol = nb_col_to_int(args.ncol) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
20 args.header = str2bool(args.header) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
21 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
22 return args |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
23 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
24 #Turn string into boolean |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
25 def str2bool(v): |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
26 if v.lower() in ('yes', 'true', 't', 'y', '1'): |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
27 return True |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
28 elif v.lower() in ('no', 'false', 'f', 'n', '0'): |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
29 return False |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
30 else: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
31 raise argparse.ArgumentTypeError('Boolean value expected.') |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
32 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
33 #return the column number in int format |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
34 def nb_col_to_int(nb_col): |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
35 try : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
36 nb_col = int(nb_col.replace("c", "")) - 1 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
37 return nb_col |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
38 except : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
39 sys.exit("Please specify the column where you would like to apply the filter with valid format") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
40 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
41 #return list of (unique) ids from string |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
42 def get_input_ids_from_string(input) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
43 ids_list = list(set(re.split(r'\s+',input.replace(";"," ").replace("\r","").replace("\n"," ").replace("\t"," ")))) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
44 if "" in ids_list : ids_list.remove("") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
45 #if "NA" in ids_list : ids_list.remove("NA") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
46 return ids_list |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
47 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
48 #return input_file and list of unique ids from input file path |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
49 def get_input_ids_from_file(input,nb_col,header) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
50 with open(input, "r") as csv_file : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
51 input_file= list(csv.reader(csv_file, delimiter='\t')) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
52 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
53 input_file, ids_list = one_id_one_line(input_file,nb_col,header) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
54 if "" in ids_list : ids_list.remove("") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
55 #if "NA" in ids_list : ids_list.remove("NA") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
56 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
57 return input_file, ids_list |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
58 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
59 #return input file by adding lines when there are more than one id per line |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
60 def one_id_one_line(input_file,nb_col,header) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
61 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
62 if header : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
63 new_file = [input_file[0]] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
64 input_file = input_file[1:] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
65 else : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
66 new_file=[] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
67 ids_list=[] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
68 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
69 for line in input_file : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
70 if line != [] and set(line) != {''}: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
71 line[nb_col] = re.sub(r"\s+","",line[nb_col]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
72 if ";" in line[nb_col] : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
73 ids = line[nb_col].split(";") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
74 for id in ids : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
75 new_file.append(line[:nb_col]+[id]+line[nb_col+1:]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
76 ids_list.append(id) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
77 else : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
78 new_file.append(line) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
79 ids_list.append(line[nb_col]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
80 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
81 ids_list= list(set(ids_list)) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
82 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
83 return new_file, ids_list |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
84 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
85 #replace all blank cells to NA |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
86 def blank_to_NA(csv_file) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
87 tmp=[] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
88 for line in csv_file : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
89 line = ["NA" if cell=="" or cell==" " or cell=="NaN" or cell=="-" else cell for cell in line] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
90 tmp.append(line) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
91 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
92 return tmp |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
93 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
94 def biogrid_output_files(ids,species) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-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"]] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
96 ids_set= set(ids) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
97 ids_not_found=set([]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
98 for id in ids : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
99 if id in ppi_dict['network'] : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
100 network_file.extend(ppi_dict['network'][id]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
101 ids_set.update([interact[1] for interact in ppi_dict['network'][id]]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
102 else : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
103 ids_not_found.add(id) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
104 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-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"]] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
106 for id in ids_set: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
107 #get pathway |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
108 if id in ppi_dict['nodes']: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
109 description_pathway=";".join(ppi_dict['nodes'][id]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
110 else : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
111 description_pathway="NA" |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
112 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
113 #get gene name |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
114 if id in ppi_dict['network']: gene_name = ppi_dict['network'][id][0][2] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
115 else : gene_name="NA" |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
116 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
117 #make line |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
118 nodes_file.append([id]+[gene_name]+[id in ids]+[id not in ids_not_found]+[description_pathway]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
119 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
120 return network_file,nodes_file |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
121 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
122 def bioplex_output_files(ids,id_type,species) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-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"]] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
124 ids_set= set(ids) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
125 ids_not_found=set([]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
126 for id in ids : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
127 if id in ppi_dict['network'][id_type] : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
128 network_file.extend(ppi_dict['network'][id_type][id]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
129 ids_set.update([interact[1] for interact in ppi_dict['network'][id_type][id]]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
130 else : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
131 ids_not_found.add(id) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
132 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-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"]] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
134 else: nodes_file=[[id_type,"Official symbol Interactor","Present in user input ids","Present in interactome","Pathway"]] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
135 for id in ids_set: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
136 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
137 if id in ppi_dict['nodes'][id_type]: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
138 description_pathway=";".join(ppi_dict['nodes'][id_type][id]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
139 else : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
140 description_pathway="NA" |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
141 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
142 #make line |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
143 if id_type=="UniProt-AC": |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
144 nodes_file.append([id]+[id in ids]+[id not in ids_not_found]+[description_pathway]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
145 elif id_type=="GeneID": |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
146 #get gene_name |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
147 if id in ppi_dict['network'][id_type]: gene_name = ppi_dict['network'][id_type][id][0][2] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
148 else : gene_name="NA" |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
149 nodes_file.append([id]+[gene_name]+[id in ids]+[id not in ids_not_found]+[description_pathway]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
150 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
151 return network_file,nodes_file |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
152 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
153 def humap_output_files(ids,species) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-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"]] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
155 ids_set= set(ids) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
156 ids_not_found=set([]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
157 for id in ids : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
158 if id in ppi_dict['network'] : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
159 network_file.extend(ppi_dict['network'][id]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
160 ids_set.update([interact[1] for interact in ppi_dict['network'][id]]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
161 else : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
162 ids_not_found.add(id) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
163 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-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"]] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
165 for id in ids_set: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
166 if id in ppi_dict['nodes']: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
167 description_pathway=";".join(ppi_dict['nodes'][id]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
168 else : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
169 description_pathway="NA" |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
170 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
171 #get gene name |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
172 if id in ppi_dict['gene_name']: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
173 gene_name = ppi_dict['gene_name'][id] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
174 else : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
175 gene_name = "NA" |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
176 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
177 #make line |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
178 nodes_file.append([id]+[gene_name]+[id in ids]+[id not in ids_not_found]+[description_pathway]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
179 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
180 return network_file,nodes_file |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
181 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
182 #function to sort the csv_file by value in a specific column |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
183 def sort_by_column(tab,sort_col,reverse,header): |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
184 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
185 if len(tab) > 1 : #if there's more than just a header or 1 row |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
186 if header : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
187 head=tab[0] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
188 tab=tab[1:] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
189 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
190 #list of empty cells in the column to sort |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
191 unsortable_lines = [i for i,line in enumerate(tab) if (line[sort_col]=='' or line[sort_col] == 'NA')] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
192 unsorted_tab=[ tab[i] for i in unsortable_lines] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
193 tab= [line for i,line in enumerate(tab) if i not in unsortable_lines] |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
194 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
195 if only_number(tab,sort_col) and any_float(tab,sort_col) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
196 tab = sorted(tab, key=lambda row: float(row[sort_col]), reverse=reverse) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
197 elif only_number(tab,sort_col): |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
198 tab = sorted(tab, key=lambda row: int(row[sort_col]), reverse=reverse) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
199 else : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
200 tab = sorted(tab, key=lambda row: row[sort_col], reverse=reverse) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
201 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
202 tab.extend(unsorted_tab) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
203 if header is True : tab = [head]+tab |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
204 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
205 return tab |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
206 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
207 def only_number(tab,col) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
208 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
209 for line in tab : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
210 if not (is_number("float",line[col].replace(",",".")) or is_number("int",line[col].replace(",","."))) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
211 return False |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
212 return True |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
213 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
214 #Check if a variable is a float or an integer |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
215 def is_number(number_format, n): |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
216 float_format = re.compile(r"^[-]?[0-9][0-9]*.?[0-9]+$") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
217 int_format = re.compile(r"^[-]?[0-9][0-9]*$") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
218 test = "" |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
219 if number_format == "int": |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
220 test = re.match(int_format, n) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
221 elif number_format == "float": |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
222 test = re.match(float_format, n) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
223 if test: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
224 return True |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
225 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
226 #return True is there is at least one float in the column |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
227 def any_float(tab,col) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
228 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
229 for line in tab : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
230 if is_number("float",line[col].replace(",",".")) : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
231 return True |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
232 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
233 return False |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
234 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
235 def main() : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
236 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
237 #Get args from command line |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
238 global args |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
239 args = get_args() |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
240 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
241 #get PPI dictionary |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
242 with open(args.dict_path, 'r') as handle: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
243 global ppi_dict |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
244 ppi_dict = json.load(handle) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
245 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
246 #Get file and/or ids from input |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
247 if args.input_type == "text" : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
248 ids = get_input_ids_from_string(args.input) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
249 elif args.input_type == "file" : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
250 input_file, ids = get_input_ids_from_file(args.input,args.ncol,args.header) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
251 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
252 #create output files |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
253 if args.database=="biogrid": |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
254 network_file, nodes_file = biogrid_output_files(ids,args.species) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
255 elif args.database=="bioplex": |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
256 network_file, nodes_file = bioplex_output_files(ids,args.id_type,args.species) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
257 elif args.database=="humap": |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
258 network_file, nodes_file = humap_output_files(ids,args.species) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
259 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
260 #convert blank to NA and sort files |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
261 network_file = blank_to_NA(network_file) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
262 network_file = sort_by_column(network_file,0,False,True) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
263 nodes_file = sort_by_column(nodes_file,0,False,True) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
264 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
265 #write output files |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
266 with open(args.network_output,"w") as output : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
267 writer = csv.writer(output,delimiter="\t") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
268 writer.writerows(network_file) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
269 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
270 with open(args.nodes_output,"w") as output : |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
271 writer = csv.writer(output,delimiter="\t") |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
272 for row in nodes_file: |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
273 writer.writerow([unicode(s).encode("utf-8") for s in row]) |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
274 |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
275 if __name__ == "__main__": |
b0ac71686b99
planemo upload commit 968cd5b4f78f0a1da86fc3bc29f8159f86e199aa-dirty
proteore
parents:
diff
changeset
|
276 main() |