annotate subtract_query.py @ 2:ff3901618482 draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
author devteam
date Thu, 22 Jun 2017 18:53:15 -0400
parents 960d5d114184
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
1 #!/usr/bin/env python
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
2 # Greg Von Kuster
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
3 """
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
4 Subtract an entire query from another query
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
5 usage: %prog in_file_1 in_file_2 begin_col end_col output
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
6 --ignore-empty-end-cols: ignore empty end columns when subtracting
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
7 """
2
ff3901618482 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents: 1
diff changeset
8 from __future__ import print_function
ff3901618482 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents: 1
diff changeset
9
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
10 import sys
2
ff3901618482 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents: 1
diff changeset
11
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
12 from bx.cookbook import doc_optparse
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
13
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
14 # Older py compatibility
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
15 try:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
16 set()
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
17 except:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
18 from sets import Set as set
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
19
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
20 assert sys.version_info[:2] >= ( 2, 4 )
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
21
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
22
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
23 def get_lines(fname, begin_col='', end_col='', ignore_empty_end_cols=False):
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
24 lines = set([])
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
25 i = 0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
26 for i, line in enumerate(file(fname)):
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
27 line = line.rstrip('\r\n')
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
28 if line and not line.startswith('#'):
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
29 if begin_col and end_col:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
30 """Both begin_col and end_col must be integers at this point."""
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
31 try:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
32 line = line.split('\t')
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
33 line = '\t'.join([line[j] for j in range(begin_col - 1, end_col)])
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
34 if ignore_empty_end_cols:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
35 # removing empty fields, we do not compare empty fields at the end of a line.
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
36 line = line.rstrip()
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
37 lines.add( line )
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
38 except:
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
39 pass
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
40 else:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
41 if ignore_empty_end_cols:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
42 # removing empty fields, we do not compare empty fields at the end of a line.
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
43 line = line.rstrip()
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
44 lines.add( line )
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
45 if i:
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
46 return (i + 1, lines)
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
47 else:
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
48 return (i, lines)
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
49
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
50
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
51 def main():
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
52 # Parsing Command Line here
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
53 options, args = doc_optparse.parse( __doc__ )
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
54
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
55 try:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
56 inp1_file, inp2_file, begin_col, end_col, out_file = args
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
57 except:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
58 doc_optparse.exception()
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
59
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
60 begin_col = begin_col.strip()
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
61 end_col = end_col.strip()
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
62
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
63 if begin_col != 'None' or end_col != 'None':
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
64 """
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
65 The user selected columns for restriction. We'll allow default
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
66 values for both begin_col and end_col as long as the user selected
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
67 at least one of them for restriction.
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
68 """
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
69 if begin_col == 'None':
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
70 begin_col = end_col
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
71 elif end_col == 'None':
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
72 end_col = begin_col
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
73 begin_col = int(begin_col)
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
74 end_col = int(end_col)
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
75 """Make sure that begin_col <= end_col (switch if not)"""
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
76 if begin_col > end_col:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
77 tmp_col = end_col
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
78 end_col = begin_col
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
79 begin_col = tmp_col
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
80 else:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
81 begin_col = end_col = ''
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
82
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
83 try:
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
84 fo = open(out_file, 'w')
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
85 except:
2
ff3901618482 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents: 1
diff changeset
86 print("Unable to open output file", file=sys.stderr)
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
87 sys.exit()
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
88
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
89 """
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
90 len1 is the number of lines in inp1_file
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
91 lines1 is the set of unique lines in inp1_file
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
92 diff1 is the number of duplicate lines removed from inp1_file
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
93 """
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
94 len1, lines1 = get_lines(inp1_file, begin_col, end_col, options.ignore_empty_end_cols)
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
95 diff1 = len1 - len(lines1)
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
96 len2, lines2 = get_lines(inp2_file, begin_col, end_col, options.ignore_empty_end_cols)
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
97
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
98 lines1.difference_update(lines2)
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
99 """lines1 is now the set of unique lines in inp1_file - the set of unique lines in inp2_file"""
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
100
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
101 for line in lines1:
2
ff3901618482 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents: 1
diff changeset
102 print(line, file=fo)
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
103
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
104 fo.close()
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
105
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
106 info_msg = 'Subtracted %d lines. ' % ((len1 - diff1) - len(lines1))
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
107
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
108 if begin_col and end_col:
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
109 info_msg += 'Restricted to columns c' + str(begin_col) + ' thru c' + str(end_col) + '. '
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
110
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
111 if diff1 > 0:
1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
112 info_msg += 'Eliminated %d duplicate/blank/comment/invalid lines from first query.' % diff1
960d5d114184 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit a1517c9d22029095120643bbe2c8fa53754dd2b7
devteam
parents: 0
diff changeset
113
2
ff3901618482 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents: 1
diff changeset
114 print(info_msg)
ff3901618482 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents: 1
diff changeset
115
0
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
116
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
117 if __name__ == "__main__":
5f6ebef89722 Imported from capsule None
devteam
parents:
diff changeset
118 main()