Mercurial > repos > proteore > proteore_venn_diagram
annotate venn_diagram.py @ 7:98b7912a9ceb draft
planemo upload commit 89cd1b5f2ca7b135abfe2d9ed18bf2478a59e099-dirty
author | proteore |
---|---|
date | Tue, 05 Mar 2019 11:25:31 -0500 |
parents | d1fd04dcb13a |
children | bca31ac678f1 |
rev | line source |
---|---|
0
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
1 #!/usr/bin/env python2.7 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
2 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
3 import os |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
4 import sys |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
5 import json |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
6 import operator |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
7 import argparse |
6
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
8 import re, csv |
0
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
9 from itertools import combinations |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
10 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
11 CURRENT_DIR = os.path.dirname(os.path.abspath(__file__)) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
12 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
13 ################################################################################################################################################## |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
14 # FUNCTIONS |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
15 ################################################################################################################################################## |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
16 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
17 def isnumber(format, n): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
18 """ |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
19 Check if an element is integer or float |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
20 """ |
3
145f347dc0e1
planemo upload commit b5beb4f7f3c60838b88a4cc32dfcb94c399d20ce-dirty
proteore
parents:
0
diff
changeset
|
21 float_format = re.compile(r"^[-]?[1-9][0-9]*.?[0-9]+$") |
145f347dc0e1
planemo upload commit b5beb4f7f3c60838b88a4cc32dfcb94c399d20ce-dirty
proteore
parents:
0
diff
changeset
|
22 int_format = re.compile(r"^[-]?[1-9][0-9]*$") |
0
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
23 test = "" |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
24 if format == "int": |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
25 test = re.match(int_format, n) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
26 elif format == "float": |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
27 test = re.match(float_format, n) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
28 if test: |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
29 return True |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
30 else: |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
31 return False |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
32 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
33 def input_to_dict(inputs): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
34 """ |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
35 Parse input and return a dictionary of name and data of each lists/files |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
36 """ |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
37 comp_dict = {} |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
38 title_dict = {} |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
39 c = ["A", "B", "C", "D", "E", "F"] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
40 for i in range(len(inputs)): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
41 input_file = inputs[i][0] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
42 name = inputs[i][1] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
43 input_type = inputs[i][2] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
44 title = c[i] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
45 title_dict[title] = name |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
46 ids = set() |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
47 if input_type == "file": |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
48 header = inputs[i][3] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
49 ncol = inputs[i][4] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
50 file_content = open(input_file, "r").readlines() |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
51 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
52 # Check if column number is in right form |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
53 if isnumber("int", ncol.replace("c", "")): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
54 if header == "true": |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
55 file_content = [x.strip() for x in [line.split("\t")[int(ncol.replace("c", ""))-1].split(";")[0] for line in file_content[1:]]] # take only first IDs |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
56 else: |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
57 file_content = [x.strip() for x in [line.split("\t")[int(ncol.replace("c", ""))-1].split(";")[0] for line in file_content]] # take only first IDs |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
58 else: |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
59 raise ValueError("Please fill in the right format of column number") |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
60 else: |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
61 ids = set() |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
62 file_content = inputs[i][0].split() |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
63 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
64 ids.update(file_content) |
6
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
65 if 'NA' in ids : ids.remove('NA') |
0
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
66 comp_dict[title] = ids |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
67 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
68 return comp_dict, title_dict |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
69 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
70 def intersect(comp_dict): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
71 """ |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
72 Calculate the intersections of input |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
73 """ |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
74 names = set(comp_dict) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
75 for i in range(1, len(comp_dict) + 1): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
76 for group in combinations(sorted(comp_dict), i): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
77 others = set() |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
78 [others.add(name) for name in names if name not in group] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
79 difference = [] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
80 intersected = set.intersection(*(comp_dict[k] for k in group)) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
81 if len(others) > 0: |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
82 difference = intersected.difference(set.union(*(comp_dict[k] for k in others))) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
83 yield group, list(intersected), list(difference) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
84 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
85 def diagram(comp_dict, title_dict): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
86 """ |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
87 Create json string for jvenn diagram plot |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
88 """ |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
89 result = {} |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
90 result["name"] = {} |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
91 for k in comp_dict.keys(): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
92 result["name"][k] = title_dict[k] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
93 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
94 result["data"] = {} |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
95 result["values"] = {} |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
96 for group, intersected, difference in intersect(comp_dict): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
97 if len(group) == 1: |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
98 result["data"]["".join(group)] = difference |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
99 result["values"]["".join(group)] = len(difference) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
100 elif len(group) > 1 and len(group) < len(comp_dict): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
101 result["data"]["".join(group)] = difference |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
102 result["values"]["".join(group)] = len(difference) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
103 elif len(group) == len(comp_dict): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
104 result["data"]["".join(group)] = intersected |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
105 result["values"]["".join(group)] = len(intersected) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
106 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
107 return result |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
108 |
6
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
109 #Write intersections of input to text output file |
0
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
110 def write_text_venn(json_result): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
111 lines = [] |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
112 result = dict((k, v) for k, v in json_result["data"].iteritems() if v != []) |
6
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
113 for key in result : |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
114 if 'NA' in result[key] : result[key].remove("NA") |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
115 list_names = dict((k, v) for k, v in json_result["name"].iteritems() if v != []) |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
116 nb_lines_max = max(len(v) for v in result.values()) |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
117 |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
118 #get list names associated to each column |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
119 column_dict = {} |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
120 for key in result : |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
121 if key in list_names : |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
122 column_dict[key] = list_names[key] |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
123 else : |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
124 keys= list(key) |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
125 column_dict[key] = "_".join([list_names[k] for k in keys]) |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
126 |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
127 #construct tsv |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
128 for key in result : |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
129 line = [column_dict[key]] |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
130 line.extend(result[key]) |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
131 if len(line) < nb_lines_max : |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
132 line.extend(['NA']*(nb_lines_max-len(line))) |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
133 lines.append(line) |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
134 #transpose tsv |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
135 lines=zip(*lines) |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
136 |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
137 with open("venn_diagram_text_output.tsv", "w") as output: |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
138 tsv_output = csv.writer(output, delimiter='\t') |
d1fd04dcb13a
planemo upload commit ad5f1c5a1a71d7fa2bc8bac408856aa80b0fc2a3
proteore
parents:
3
diff
changeset
|
139 tsv_output.writerows(lines) |
0
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
140 |
3
145f347dc0e1
planemo upload commit b5beb4f7f3c60838b88a4cc32dfcb94c399d20ce-dirty
proteore
parents:
0
diff
changeset
|
141 def write_summary(summary_file, inputs): |
0
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
142 """ |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
143 Paste json string into template file |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
144 """ |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
145 a, b = input_to_dict(inputs) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
146 data = diagram(a, b) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
147 write_text_venn(data) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
148 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
149 to_replace = { |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
150 "series": [data], |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
151 "displayStat": "true", |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
152 "displaySwitch": "true", |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
153 "shortNumber": "true", |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
154 } |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
155 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
156 FH_summary_tpl = open(os.path.join(CURRENT_DIR, "jvenn_template.html")) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
157 FH_summary_out = open(summary_file, "w" ) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
158 for line in FH_summary_tpl: |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
159 if "###JVENN_DATA###" in line: |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
160 line = line.replace("###JVENN_DATA###", json.dumps(to_replace)) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
161 FH_summary_out.write(line) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
162 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
163 FH_summary_out.close() |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
164 FH_summary_tpl.close() |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
165 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
166 def process(args): |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
167 write_summary(args.summary, args.input) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
168 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
169 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
170 ################################################################################################################################################## |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
171 # MAIN |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
172 ################################################################################################################################################## |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
173 if __name__ == '__main__': |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
174 # Parse parameters |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
175 parser = argparse.ArgumentParser(description='Filters an abundance file') |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
176 group_input = parser.add_argument_group( 'Inputs' ) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
177 group_input.add_argument('--input', nargs="+", action="append", required=True, help="The input tabular file.") |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
178 group_output = parser.add_argument_group( 'Outputs' ) |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
179 group_output.add_argument('--summary', default="summary.html", help="The HTML file containing the graphs. [Default: %(default)s]") |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
180 args = parser.parse_args() |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
181 |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
182 # Process |
9d4f20618ab4
planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff
changeset
|
183 process( args ) |