Mercurial > repos > proteore > proteore_reactome

annotate reactome_analysis.py @ 11:19195d1a4063 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
author proteore
date Mon, 10 May 2021 15:30:34 +0000
parents a58dc5d4b8cd
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
1 import argparse
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
2 import csv
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
3 import json
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
4 import os
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
5 import re
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
6
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
7 CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
8
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
9
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
10 def id_valid(identifiers):
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
11 """
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
12 Validate IDs if they contain special characters
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
13 """
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
14 res = []
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
15 remove = []
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
16 for id in identifiers:
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
17 id = id.split(";")[0]
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
18 if re.match("^[A-Za-z0-9_-]*$", id):
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
19 res.append(id)
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
20 else:
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
21 remove.append(id)
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
22 return res, remove
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
23
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
24
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
25 def isnumber(format, n):
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
26 """
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
27 Check if an variable is numeric
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
28 """
4
878128362e33 planemo upload commit 34edf9c3db61650a03d5a0d548b9697a94ecde34-dirty
proteore
parents: 0
diff changeset
29 float_format = re.compile(r"^[-]?[1-9][0-9]*.?[0-9]+$")
878128362e33 planemo upload commit 34edf9c3db61650a03d5a0d548b9697a94ecde34-dirty
proteore
parents: 0
diff changeset
30 int_format = re.compile(r"^[-]?[1-9][0-9]*$")
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
31 test = ""
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
32 if format == "int":
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
33 test = re.match(int_format, n)
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
34 elif format == "float":
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
35 test = re.match(float_format, n)
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
36 if test:
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
37 return True
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
38 else:
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
39 return False
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
40
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
41
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
42 def data_json(identifiers):
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
43 """
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
44 Submit IDs list to Reactome and return results in json format
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
45 Return error in HTML format if web service is not available
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
46 """
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
47 trash = []
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
48 if identifiers[1] == "list":
8
a58dc5d4b8cd planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
49 ids = identifiers[0].split()
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
50 ids = [x.split(";") for x in ids]
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
51 ids = [item.strip() for sublist in ids for item in sublist if item != ''] # noqa 501
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
52 json_string = os.popen("curl -H \"Content-Type: text/plain\" -d \"$(printf '%s')\" -X POST --url www.reactome.org/AnalysisService/identifiers/\?pageSize\=1\&page\=1" % ids).read() # noqa 501
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
53 if len(id_valid(identifiers[0].split())[1]) > 0:
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
54 trash = id_valid(identifiers[0].split())[1]
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
55 elif identifiers[1] == "file":
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
56 header = identifiers[2]
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
57 with open(identifiers[0], "r") as mq:
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
58 file_content = csv.reader(mq, delimiter="\t")
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
59 file_content = list(file_content) # csv object to list
8
a58dc5d4b8cd planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
60 ncol = identifiers[3]
a58dc5d4b8cd planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
61 if isnumber("int", ncol.replace("c", "")):
a58dc5d4b8cd planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
62 if header == "true":
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
63 idens = [x for x in [line[int(ncol.replace("c", ""))-1].split(";") for line in file_content[1:]]] # noqa 501
8
a58dc5d4b8cd planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
64 else:
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
65 idens = [x for x in [line[int(ncol.replace("c", ""))-1].split(";") for line in file_content]] # noqa 501
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
66 # flat list of list of lists, remove empty items
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
67 idens = [item.strip() for sublist in idens for item in sublist if item != ''] # noqa 501
8
a58dc5d4b8cd planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
68 ids = "\n".join(id_valid(idens)[0])
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
69 json_string = os.popen("curl -H \"Content-Type: text/plain\" -d \"$(printf '%s')\" -X POST --url www.reactome.org/AnalysisService/identifiers/\?pageSize\=1\&page\=1 2> stderr" % ids).read() # noqa 501
8
a58dc5d4b8cd planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
70 if len(id_valid(idens)[1]) > 0:
a58dc5d4b8cd planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
71 trash = id_valid(idens)[1]
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
72 # print(json_string)
6
9cc475dcd0f2 planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents: 4
diff changeset
73 j = json.loads(json_string)
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
74 print("Identifiers not found: " + str(j["identifiersNotFound"]))
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
75 print("Pathways found: " + str(j["pathwaysFound"]))
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
76 return json_string, trash
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
77
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
78
6
9cc475dcd0f2 planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents: 4
diff changeset
79 def write_output(filename, json_string, species, trash_file, trash):
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
80 """
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
81 Replace json result in template and print to output
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
82 """
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
83 template = open(os.path.join(CURRENT_DIR, "template.html"))
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
84 output = open(filename, "w")
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
85 try:
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
86 for line in template:
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
87 if "{token}" in line:
6
9cc475dcd0f2 planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents: 4
diff changeset
88 line = line.replace("{species}", species)
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
89 line = line.replace("{token}", json.loads(json_string)["summary"]["token"]) # noqa 501
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
90 output.write(line)
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
91 except ValueError:
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
92 output.write("An error occurred due to unavailability of Reactome web service. Please return later.") # noqa 501
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
93 template.close()
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
94 output.close()
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
95
4
878128362e33 planemo upload commit 34edf9c3db61650a03d5a0d548b9697a94ecde34-dirty
proteore
parents: 0
diff changeset
96 if trash:
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
97 # print(trash)
4
878128362e33 planemo upload commit 34edf9c3db61650a03d5a0d548b9697a94ecde34-dirty
proteore
parents: 0
diff changeset
98 trash_out = open(trash_file, "w")
878128362e33 planemo upload commit 34edf9c3db61650a03d5a0d548b9697a94ecde34-dirty
proteore
parents: 0
diff changeset
99 trash_out.write("\n".join(trash))
878128362e33 planemo upload commit 34edf9c3db61650a03d5a0d548b9697a94ecde34-dirty
proteore
parents: 0
diff changeset
100 trash_out.close()
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
101
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
102
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
103 def options():
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
104 parser = argparse.ArgumentParser()
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
105 argument = parser.add_argument("--json", nargs="+", required=True)
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
106 argument = parser.add_argument("--output", default="output.html")
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
107 argument = parser.add_argument("--trash", default="trash.txt")
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
108 argument = parser.add_argument("--species", default="48887") # noqa 841
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
109 args = parser.parse_args()
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
110 filename = args.output
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
111 json_string, trash = data_json(args.json)
6
9cc475dcd0f2 planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents: 4
diff changeset
112 write_output(filename, json_string, args.species, args.trash, trash)
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
113
11
19195d1a4063 "planemo upload commit a2de8dd8f2468046e787f83812d201bb191edab8-dirty"
proteore
parents: 8
diff changeset
114
0
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
115 if __name__ == "__main__":
216bd2a75b1d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
116 options()