comparison venner.py @ 0:8ea9b4e5a389

Uploaded
author g2cmnty@test-web1.g2.bx.psu.edu
date Wed, 22 Jun 2011 03:28:25 -0400
parents
children cc6707a1e044
comparison
equal deleted inserted replaced
-1:000000000000 0:8ea9b4e5a389
1 #!/usr/bin/env python
2 '''
3 Created on Jul 13, 2010
4
5 Generates Venn diagram from 2 or three input files
6
7 one could generate output file that for each row contain the segment
8 i.e. A, B, AB, AmB, BmA
9
10 @author: Ido M. Tamir
11 '''
12 from mako.template import Template
13 from optparse import OptionParser
14 import urllib,re,sys
15
16 class Bunch:
17 def __init__(self, d):
18 for k, v in d.items():
19 if isinstance(v, dict):
20 v = Bunch(v)
21 self.__dict__[k] = v
22
23 class VennFile():
24 def __init__(self, filePath, column, name):
25 self.filePath = filePath
26 self.column = column
27 cleanname = re.sub("/","",name)
28 self.name = urllib.quote(cleanname, safe="%/:=&?~#+!$,;'@()*[]")
29 self.dict = {}
30 def read(self):
31 dict = {}
32 lineNr = 0
33 for line in open( self.filePath, 'rb'):
34 key = line.split("\t")[self.column].strip()
35 linesList = dict.get(key, [])
36 linesList.append(line+"\t"+str(lineNr))
37 dict[key] = linesList
38 lineNr += 1
39 self.dict = dict
40 return self
41
42 class Venn2:
43 def __init__(self, title, size, one, two):
44 self.one = one.read()
45 self.two = two.read()
46 self.title = title
47 self.size = size
48
49 def toUrl(self):
50 one_keys = set(self.one.dict.keys())
51 two_keys = set(self.two.dict.keys())
52
53 keys_one_i_two = one_keys.intersection(two_keys)
54
55 total = len(one_keys) + len(two_keys)
56 sizes = [len(one_keys), len(two_keys), 0, len(keys_one_i_two)]
57 sizes = self.relSizes(sizes, total)
58 names = [self.one.name, self.two.name]
59 return self.url(total, sizes, names)
60
61 def relSizes(self, sizes, total):
62 return map(lambda s: str(int(round((s/float(total) * 100)))), sizes)
63
64 def url(self, total, sizes, names):
65 base = "http://chart.apis.google.com/chart?cht=v&chd=t:"
66 counts = ",".join(sizes)
67 titlep = "&chtt="+self.title
68 size = "&chs="+str(self.size)+"x"+str(self.size)
69 legend = "&chdl="+"|".join(names)
70 url = base+counts+titlep+size+legend
71 return url
72
73 def toHtml(self):
74 one_keys = set(self.one.dict.keys())
75 two_keys = set(self.two.dict.keys())
76
77 numbers = Bunch({
78 "one_keys" : len(set(self.one.dict.keys())),
79 "two_keys" : len(set(self.two.dict.keys())),
80 "one_only" : len(one_keys.difference(two_keys)),
81 "two_only" : len(two_keys.difference(one_keys)),
82 "one_i_two" : len(one_keys.intersection(two_keys)),
83 })
84
85 template = """
86 <html>
87 <head>
88 <title>Venn diagram ${title}</title>
89 </head>
90 <body>
91 <h3>${ title }</h3>
92 <div>
93 <img src="${ url }"/>
94 </div>
95 <div>
96 <table>
97 <tr><th>Segment</th><th>Count</th></tr>
98 <tr><td>${ one }</td><td>${ n.one_keys }</td></tr>
99 <tr><td>${ two }</td><td>${ n.two_keys }</td></tr>
100 <tr><td>${ one } \ ${ two }</td><td>${ n.one_only }</td></tr>
101 <tr><td>${ two } \ ${ one }</td><td>${ n.two_only }</td></tr>
102 <tr><td>${ one } &cap; ${ two }</td><td>${ n.one_i_two }</td></tr>
103 </table>
104 </div>
105 </body>
106 </html>"""
107 result = Template(template).render(one=self.one.name, two=self.two.name, n=numbers, title=self.title, url=self.toUrl())
108 return(result)
109
110
111
112 class Venn3(Venn2):
113 def __init__(self, title, size, one, two, three):
114 Venn2.__init__(self, title, size, one, two)
115 self.three = three.read()
116
117 def toUrl(self):
118 one_keys = set(self.one.dict.keys())
119 two_keys = set(self.two.dict.keys())
120 three_keys = set(self.three.dict.keys())
121
122 keys_one_i_two = one_keys.intersection(two_keys)
123 keys_one_i_three = one_keys.intersection(three_keys)
124 keys_two_i_three = two_keys.intersection(three_keys)
125 keys_one_i_two_i_three = one_keys.intersection(two_keys).intersection(three_keys)
126
127 total = len(one_keys)+len(two_keys)+len(three_keys)
128 sizes = [len(one_keys), len(two_keys), len(three_keys), len(keys_one_i_two), len(keys_one_i_three), len(keys_two_i_three), len(keys_one_i_two_i_three)]
129 sizes = self.relSizes(sizes, total)
130 names = [self.one.name, self.two.name, self.three.name]
131 return self.url(total, sizes, names)
132
133 def toHtml(self):
134 one_keys = set(self.one.dict.keys())
135 two_keys = set(self.two.dict.keys())
136 three_keys = set(self.three.dict.keys())
137
138 xa = one_keys.intersection(two_keys)
139 xt = two_keys.intersection(three_keys)
140 xd = xt.difference(one_keys)
141
142 numbers = Bunch({
143 "one_keys" : len(set(self.one.dict.keys())),
144 "two_keys" : len(set(self.two.dict.keys())),
145 "three_keys" : len(set(self.three.dict.keys())),
146 "one_only" : len(one_keys.difference(two_keys.union(three_keys))),
147 "two_only" : len(two_keys.difference(one_keys.union(three_keys))),
148 "three_only" : len(three_keys.difference(one_keys.union(two_keys))),
149 "one_two" : len(one_keys.intersection(two_keys).difference(three_keys)),
150 "one_three" : len(one_keys.intersection(three_keys).difference(two_keys)),
151 "two_three" : len(two_keys.intersection(three_keys).difference(one_keys)),
152 "one_i_two_i_three" : len(one_keys.intersection(two_keys).intersection(three_keys))
153 })
154
155 template = """
156 <html>
157 <head>
158 <title>Venn diagram ${title}</title>
159 </head>
160 <body>
161 <h3>${ title }</h3>
162 <div>
163 <img src="${ url }"/>
164 </div>
165 <div>
166 <table>
167 <tr><th>Segment</th><th>Count</th></tr>
168 <tr><td>${ one }</td><td>${ n.one_keys }</td></tr>
169 <tr><td>${ two }</td><td>${ n.two_keys }</td></tr>
170 <tr><td>${ three }</td><td>${ n.three_keys }</td></tr>
171 <tr><td>${ one } \ (${ two } &cup; ${ three })</td><td>${ n.one_only }</td></tr>
172 <tr><td>${ two } \ (${ one } &cup; ${ three})</td><td>${ n.two_only }</td></tr>
173 <tr><td>${ three } \ (${ one } &cup; ${ two })</td><td>${ n.three_only }</td></tr>
174 <tr><td>${ one } &cap; ${ two } \ ${ three } </td><td>${ n.one_two }</td></tr>
175 <tr><td>${ one } &cap; ${ three } \ ${ two } </td><td>${ n.one_three }</td></tr>
176 <tr><td>${ two } &cap; ${ three } \ ${ one } </td><td>${ n.two_three }</td></tr>
177 <tr><td>${ one } &cap; ${ two } &cap; ${ three }</td><td>${ n.one_i_two_i_three }</td></tr>
178 </table>
179 </div>
180 </body>
181 </html>"""
182 result = Template(template).render(one=self.one.name, two=self.two.name, three=self.three.name, n=numbers, title=self.title, url=self.toUrl())
183 return(result)
184
185
186
187
188 def main():
189 '''main worker func'''
190 parser = OptionParser()
191 parser.add_option( "--files", dest="filePaths", help="file paths delimited by ,")
192 parser.add_option( "--columns", dest="columns", help="0 based columnIndices delimited by ,")
193 parser.add_option( "--asNames", dest="asNames", help="names of the columns for pretty print")
194 parser.add_option( "--title", dest="title", help="title of plot")
195 parser.add_option( "--size", dest="size", help="size plot, default 300")
196 parser.add_option( "--outname", dest="outfileHtml", help="path of generated html file")
197
198 (o, args) = parser.parse_args()
199 errors = []
200 if o.filePaths is None:
201 errors.append("please add required paths to files")
202 if o.columns is None:
203 errors.append( "please add required columns" )
204 if o.asNames is None:
205 errors.append( "please add required asNames")
206 if len(errors) > 0:
207 print("\n".join(errors))
208 sys.exit()
209 filePaths = o.filePaths.split(",")
210 columns = o.columns.split(",")
211 columns = map(int, columns)
212 asNames = o.asNames.split(",")
213 if len(errors) > 0 and ( len(filePaths) != len(columns) or len(columns) != len(asNames) ):
214 errors.append( "different length of filePaths, columns or names:" +o.columns+" "+" "+o.names+" "+o.filePaths )
215 title = ""
216 if o.title:
217 title = o.title
218 if o.outfileHtml is None:
219 errors.append( "please add outfile name for html" )
220 if len(filePaths) > 3:
221 errors.append( "can only compare up to three files was:"+str(len(filePaths)))
222 if len(filePaths) == 1:
223 errors.append( "just one file to compare does not make sense!")
224 if len(errors) > 0:
225 print("\n".join(errors))
226 sys.exit()
227
228 size = "300"
229 if o.size:
230 size = o.size
231
232 fileCount = len(filePaths)
233 if fileCount == 2:
234 venn = Venn2(title, size, VennFile(filePaths[0],columns[0],asNames[0]), VennFile(filePaths[1], columns[1], asNames[1]))
235 else:
236 venn = Venn3(title, size, VennFile(filePaths[0],columns[0],asNames[0]), VennFile(filePaths[1], columns[1], asNames[1]), VennFile(filePaths[2],columns[2],asNames[2]))
237 htmlText = venn.toHtml()
238 html = open(o.outfileHtml, 'w')
239 try:
240 html.write(htmlText)
241 finally:
242 html.close()
243
244
245
246
247 if __name__ == '__main__':
248 main()
249
250 #$ python venner.py --files testFiles/fileA.tab,testFiles/fileB.tab --columns 1,1 --outname out.html --asNames As,Bs
251