Mercurial > repos > galaxyp > proteore_venn_diagram

annotate venn_diagram.py @ 0:57f01ca855cd draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
author galaxyp
date Sat, 12 Jun 2021 18:06:28 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
1 #!/usr/bin/env python
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
2
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
3 import argparse
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
4 import csv
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
5 import json
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
6 import os
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
7 import re
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
8 from itertools import combinations
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
9
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
10
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
11 CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
12
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
13 ########################################################################
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
14 # FUNCTIONS
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
15 ########################################################################
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
16
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
17
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
18 def isnumber(format, n):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
19 """
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
20 Check if an element is integer or float
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
21 """
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
22 float_format = re.compile(r"^[-]?[1-9][0-9]*.?[0-9]+$")
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
23 int_format = re.compile(r"^[-]?[1-9][0-9]*$")
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
24 test = ""
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
25 if format == "int":
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
26 test = re.match(int_format, n)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
27 elif format == "float":
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
28 test = re.match(float_format, n)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
29 if test:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
30 return True
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
31 else:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
32 return False
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
33
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
34
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
35 def input_to_dict(inputs):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
36 """
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
37 Parse input and return a dictionary of name and data of each lists/files
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
38 """
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
39 comp_dict = {}
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
40 title_dict = {}
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
41 c = ["A", "B", "C", "D", "E", "F"]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
42 for i in range(len(inputs)):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
43 input_file = inputs[i][0]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
44 name = inputs[i][1]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
45 input_type = inputs[i][2]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
46 title = c[i]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
47 title_dict[title] = name
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
48 ids = set()
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
49 if input_type == "file":
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
50 header = inputs[i][3]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
51 ncol = inputs[i][4]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
52 with open(input_file, "r") as handle:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
53 file_content = csv.reader(handle, delimiter="\t")
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
54 file_content = list(file_content) # csv object to list
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
55
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
56 # Check if column number is in right form
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
57 if isnumber("int", ncol.replace("c", "")):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
58 if header == "true":
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
59 # gets ids from defined column
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
60 file_content = [x for x in [line[int(ncol.replace("c", ""))-1].split(";") for line in file_content[1:]]] # noqa 501
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
61
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
62 else:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
63 file_content = [x for x in [line[int(ncol.replace("c", ""))-1].split(";") for line in file_content]] # noqa 501
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
64 else:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
65 raise ValueError("Please fill in the right format of column number") # noqa 501
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
66 else:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
67 ids = set()
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
68 file_content = inputs[i][0].split()
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
69 file_content = [x.split(";") for x in file_content]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
70
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
71 # flat list of list of lists, remove empty items
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
72 file_content = [item.strip() for sublist in file_content for item in sublist if item != ''] # noqa 501
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
73 ids.update(file_content)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
74 if 'NA' in ids:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
75 ids.remove('NA')
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
76 comp_dict[title] = ids
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
77
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
78 return comp_dict, title_dict
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
79
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
80
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
81 def intersect(comp_dict):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
82 """
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
83 Calculate the intersections of input
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
84 """
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
85 names = set(comp_dict)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
86 for i in range(1, len(comp_dict) + 1):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
87 for group in combinations(sorted(comp_dict), i):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
88 others = set()
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
89 [others.add(name) for name in names if name not in group]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
90 difference = []
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
91 intersected = set.intersection(*(comp_dict[k] for k in group))
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
92 if len(others) > 0:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
93 difference = intersected.difference(set.union(*(comp_dict[k] for k in others))) # noqa 501
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
94 yield group, list(intersected), list(difference)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
95
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
96
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
97 def diagram(comp_dict, title_dict):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
98 """
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
99 Create json string for jvenn diagram plot
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
100 """
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
101 result = {}
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
102 result["name"] = {}
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
103 for k in comp_dict.keys():
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
104 result["name"][k] = title_dict[k]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
105
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
106 result["data"] = {}
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
107 result["values"] = {}
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
108 for group, intersected, difference in intersect(comp_dict):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
109 if len(group) == 1:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
110 result["data"]["".join(group)] = sorted(difference)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
111 result["values"]["".join(group)] = len(difference)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
112 elif len(group) > 1 and len(group) < len(comp_dict):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
113 result["data"]["".join(group)] = sorted(difference)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
114 result["values"]["".join(group)] = len(difference)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
115 elif len(group) == len(comp_dict):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
116 result["data"]["".join(group)] = sorted(intersected)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
117 result["values"]["".join(group)] = len(intersected)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
118
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
119 return result
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
120
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
121 # Write intersections of input to text output file
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
122
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
123
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
124 def write_text_venn(json_result):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
125 lines = []
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
126 result = dict((k, v) for k, v in json_result["data"].items() if v != []) # noqa 501
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
127 for key in result:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
128 if 'NA' in result[key]:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
129 result[key].remove("NA")
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
130
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
131 list_names = dict((k, v) for k, v in json_result["name"].items() if v != []) # noqa 501
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
132 nb_lines_max = max(len(v) for v in result.values())
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
133
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
134 # get list names associated to each column
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
135 column_dict = {}
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
136 for key in result:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
137 if key in list_names:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
138 column_dict[key] = list_names[key]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
139 else:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
140 keys = list(key)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
141 column_dict[key] = "_".join([list_names[k] for k in keys])
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
142
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
143 # construct tsv
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
144 for key in result:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
145 line = result[key]
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
146 if len(line) < nb_lines_max:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
147 line.extend([''] * (nb_lines_max - len(line)))
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
148 line = [column_dict[key]] + line # add header
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
149 lines.append(line)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
150 # transpose tsv
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
151 lines = zip(*lines)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
152
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
153 with open("venn_diagram_text_output.tsv", "w") as output:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
154 tsv_output = csv.writer(output, delimiter='\t')
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
155 tsv_output.writerows(lines)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
156
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
157
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
158 def write_summary(summary_file, inputs):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
159 """
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
160 Paste json string into template file
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
161 """
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
162 a, b = input_to_dict(inputs)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
163 data = diagram(a, b)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
164 write_text_venn(data)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
165
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
166 to_replace = {
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
167 "series": [data],
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
168 "displayStat": "true",
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
169 "displaySwitch": "true",
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
170 "shortNumber": "true",
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
171 }
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
172
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
173 FH_summary_tpl = open(os.path.join(CURRENT_DIR, "jvenn_template.html"))
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
174 FH_summary_out = open(summary_file, "w")
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
175 for line in FH_summary_tpl:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
176 if "###JVENN_DATA###" in line:
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
177 line = line.replace("###JVENN_DATA###", json.dumps(to_replace))
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
178 FH_summary_out.write(line)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
179
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
180 FH_summary_out.close()
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
181 FH_summary_tpl.close()
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
182
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
183
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
184 def process(args):
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
185 write_summary(args.summary, args.input)
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
186
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
187
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
188 #####################################################################
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
189 # MAIN
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
190 #####################################################################
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
191 if __name__ == '__main__':
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
192 # Parse parameters
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
193 parser = argparse.ArgumentParser(description='Filters an abundance file')
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
194 group_input = parser.add_argument_group('Inputs')
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
195 group_input.add_argument('--input', nargs="+", action="append",
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
196 required=True, help="The input tabular file.")
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
197 group_output = parser.add_argument_group('Outputs')
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
198 group_output.add_argument('--summary', default="summary.html",
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
199 help="The HTML file containing the graphs. \
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
200 [Default: %(default)s]")
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
201 args = parser.parse_args()
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
202
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
203 # Process
57f01ca855cd "planemo upload commit 47d779aa1de5153673ac8bb1e37c9730210cbb5d"
galaxyp
parents:
diff changeset
204 process(args)