Mercurial > repos > proteore > proteore_venn_diagram

annotate venn_diagram.py @ 6:95bbe3a6b09f draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit d78a9195d250f3d873f3c4ab67d1fa3ae58926f5-dirty"
author proteore
date Fri, 10 Jan 2020 08:18:06 +0000
parents c3bbc688dde2
children 087011d9e3aa
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
1 #!/usr/bin/env python2.7
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
2
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
3 import os
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
4 import sys
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
5 import json
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
6 import operator
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
7 import argparse
1
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
8 import re, csv
0
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
9 from itertools import combinations
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
10
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
11 CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
12
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
13 ##################################################################################################################################################
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
14 # FUNCTIONS
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
15 ##################################################################################################################################################
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
16
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
17 def isnumber(format, n):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
18 """
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
19 Check if an element is integer or float
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
20 """
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
21 float_format = re.compile(r"^[-]?[1-9][0-9]*.?[0-9]+$")
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
22 int_format = re.compile(r"^[-]?[1-9][0-9]*$")
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
23 test = ""
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
24 if format == "int":
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
25 test = re.match(int_format, n)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
26 elif format == "float":
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
27 test = re.match(float_format, n)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
28 if test:
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
29 return True
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
30 else:
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
31 return False
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
32
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
33 def input_to_dict(inputs):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
34 """
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
35 Parse input and return a dictionary of name and data of each lists/files
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
36 """
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
37 comp_dict = {}
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
38 title_dict = {}
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
39 c = ["A", "B", "C", "D", "E", "F"]
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
40 for i in range(len(inputs)):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
41 input_file = inputs[i][0]
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
42 name = inputs[i][1]
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
43 input_type = inputs[i][2]
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
44 title = c[i]
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
45 title_dict[title] = name
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
46 ids = set()
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
47 if input_type == "file":
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
48 header = inputs[i][3]
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
49 ncol = inputs[i][4]
4
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
50 with open(input_file,"r") as handle :
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
51 file_content = csv.reader(handle,delimiter="\t")
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
52 file_content = list(file_content) #csv object to list
0
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
53
4
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
54 # Check if column number is in right form
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
55 if isnumber("int", ncol.replace("c", "")):
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
56 if header == "true":
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
57 file_content = [x for x in [line[int(ncol.replace("c", ""))-1].split(";") for line in file_content[1:]]] # gets ids from defined column
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
58 else:
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
59 file_content = [x for x in [line[int(ncol.replace("c", ""))-1].split(";") for line in file_content]]
0
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
60 else:
4
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
61 raise ValueError("Please fill in the right format of column number")
0
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
62 else:
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
63 ids = set()
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
64 file_content = inputs[i][0].split()
4
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
65 file_content = [x.split(";") for x in file_content]
0
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
66
4
c3bbc688dde2 planemo upload commit 1316c2feac23e3b1a4f929c3386142d5d5d7e37a-dirty
proteore
parents: 3
diff changeset
67 file_content = [item.strip() for sublist in file_content for item in sublist if item != ''] #flat list of list of lists, remove empty items
0
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
68 ids.update(file_content)
1
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
69 if 'NA' in ids : ids.remove('NA')
0
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
70 comp_dict[title] = ids
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
71
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
72 return comp_dict, title_dict
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
73
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
74 def intersect(comp_dict):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
75 """
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
76 Calculate the intersections of input
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
77 """
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
78 names = set(comp_dict)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
79 for i in range(1, len(comp_dict) + 1):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
80 for group in combinations(sorted(comp_dict), i):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
81 others = set()
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
82 [others.add(name) for name in names if name not in group]
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
83 difference = []
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
84 intersected = set.intersection(*(comp_dict[k] for k in group))
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
85 if len(others) > 0:
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
86 difference = intersected.difference(set.union(*(comp_dict[k] for k in others)))
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
87 yield group, list(intersected), list(difference)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
88
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
89 def diagram(comp_dict, title_dict):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
90 """
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
91 Create json string for jvenn diagram plot
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
92 """
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
93 result = {}
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
94 result["name"] = {}
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
95 for k in comp_dict.keys():
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
96 result["name"][k] = title_dict[k]
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
97
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
98 result["data"] = {}
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
99 result["values"] = {}
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
100 for group, intersected, difference in intersect(comp_dict):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
101 if len(group) == 1:
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
102 result["data"]["".join(group)] = difference
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
103 result["values"]["".join(group)] = len(difference)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
104 elif len(group) > 1 and len(group) < len(comp_dict):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
105 result["data"]["".join(group)] = difference
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
106 result["values"]["".join(group)] = len(difference)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
107 elif len(group) == len(comp_dict):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
108 result["data"]["".join(group)] = intersected
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
109 result["values"]["".join(group)] = len(intersected)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
110
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
111 return result
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
112
1
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
113 #Write intersections of input to text output file
0
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
114 def write_text_venn(json_result):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
115 lines = []
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
116 result = dict((k, v) for k, v in json_result["data"].iteritems() if v != [])
1
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
117 for key in result :
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
118 if 'NA' in result[key] : result[key].remove("NA")
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
119 list_names = dict((k, v) for k, v in json_result["name"].iteritems() if v != [])
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
120 nb_lines_max = max(len(v) for v in result.values())
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
121
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
122 #get list names associated to each column
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
123 column_dict = {}
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
124 for key in result :
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
125 if key in list_names :
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
126 column_dict[key] = list_names[key]
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
127 else :
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
128 keys= list(key)
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
129 column_dict[key] = "_".join([list_names[k] for k in keys])
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
130
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
131 #construct tsv
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
132 for key in result :
6
95bbe3a6b09f "planemo upload commit d78a9195d250f3d873f3c4ab67d1fa3ae58926f5-dirty"
proteore
parents: 4
diff changeset
133 line = result[key]
1
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
134 if len(line) < nb_lines_max :
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
135 line.extend(['NA']*(nb_lines_max-len(line)))
6
95bbe3a6b09f "planemo upload commit d78a9195d250f3d873f3c4ab67d1fa3ae58926f5-dirty"
proteore
parents: 4
diff changeset
136 line = [column_dict[key]] + line #add header
1
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
137 lines.append(line)
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
138 #transpose tsv
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
139 lines=zip(*lines)
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
140
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
141 with open("venn_diagram_text_output.tsv", "w") as output:
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
142 tsv_output = csv.writer(output, delimiter='\t')
b17f0fbbd6c9 planemo upload commit ba044b029418f46b923830e8eaa1088833e6ef6b-dirty
proteore
parents: 0
diff changeset
143 tsv_output.writerows(lines)
0
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
144
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
145 def write_summary(summary_file, inputs):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
146 """
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
147 Paste json string into template file
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
148 """
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
149 a, b = input_to_dict(inputs)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
150 data = diagram(a, b)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
151 write_text_venn(data)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
152
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
153 to_replace = {
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
154 "series": [data],
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
155 "displayStat": "true",
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
156 "displaySwitch": "true",
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
157 "shortNumber": "true",
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
158 }
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
159
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
160 FH_summary_tpl = open(os.path.join(CURRENT_DIR, "jvenn_template.html"))
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
161 FH_summary_out = open(summary_file, "w" )
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
162 for line in FH_summary_tpl:
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
163 if "###JVENN_DATA###" in line:
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
164 line = line.replace("###JVENN_DATA###", json.dumps(to_replace))
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
165 FH_summary_out.write(line)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
166
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
167 FH_summary_out.close()
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
168 FH_summary_tpl.close()
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
169
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
170 def process(args):
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
171 write_summary(args.summary, args.input)
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
172
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
173
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
174 ##################################################################################################################################################
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
175 # MAIN
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
176 ##################################################################################################################################################
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
177 if __name__ == '__main__':
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
178 # Parse parameters
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
179 parser = argparse.ArgumentParser(description='Filters an abundance file')
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
180 group_input = parser.add_argument_group( 'Inputs' )
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
181 group_input.add_argument('--input', nargs="+", action="append", required=True, help="The input tabular file.")
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
182 group_output = parser.add_argument_group( 'Outputs' )
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
183 group_output.add_argument('--summary', default="summary.html", help="The HTML file containing the graphs. [Default: %(default)s]")
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
184 args = parser.parse_args()
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
185
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
186 # Process
8d61f5ca50a4 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
187 process( args )