Mercurial > repos > miller-lab > genome_diversity

annotate reorder.py @ 28:184d14e4270d

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Update to Miller Lab devshed revision 4ede22dd5500
author Richard Burhans <burhans@bx.psu.edu>
date Wed, 17 Jul 2013 12:46:46 -0400
parents 248b06e86022
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
1 #!/usr/bin/env python
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
2
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
3 import sys
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
4
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
5 def parse_rangelist(string):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
6 rv = []
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
7
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
8 tokens = strip_split(string, ',')
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
9 for token in tokens:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
10 int_list = parse_token(token)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
11 for int_val in int_list:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
12 int_val -= 1
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
13 if int_val not in rv:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
14 rv.append(int_val)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
15
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
16 return rv
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
17
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
18 def parse_token(token):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
19 values = strip_split(token, '-')
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
20 num_values = len(values)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
21
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
22 if num_values not in [1, 2]:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
23 print >> sys.stderr, 'Error: "%s" is not a valid range' % token
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
24 sys.exit(1)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
25
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
26 int_list = []
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
27 for value in values:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
28 if value:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
29 int_val = as_int(value)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
30
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
31 if int_val < 1:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
32 print >> sys.stderr, 'Error: "%s" is not >= 1' % value
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
33 sys.exit(1)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
34
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
35 int_list.append(int_val)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
36 else:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
37 print >> sys.stderr, 'Error: "%s" is not a valid range' % token
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
38 sys.exit(1)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
39
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
40 if num_values == 1:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
41 return int_list
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
42
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
43 a, b = int_list
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
44
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
45 if a <= b:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
46 return range(a, b+1)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
47 else:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
48 return range(a, b-1, -1)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
49
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
50 def strip_split(string, delim):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
51 return [elem.strip() for elem in string.split(delim)]
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
52
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
53 def as_int(string):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
54 try:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
55 val = int(string)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
56 except:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
57 print >> sys.stderr, 'Error: "%s" does not appear to be an integer' % string
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
58 sys.exit(1)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
59 return val
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
60
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
61 def get_lines(filename):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
62 rv = []
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
63
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
64 fh = open(filename)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
65 for line in fh:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
66 line = line.rstrip('\r\n')
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
67 rv.append(line)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
68 fh.close()
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
69
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
70 return rv
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
71
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
72 def reorder(old_lines, new_order, filename):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
73 max_index = len(old_lines) - 1
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
74
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
75 fh = open(filename, 'w')
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
76
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
77 for index in new_order:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
78 if index <= max_index:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
79 print >> fh, old_lines[index]
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
80 old_lines[index] = None
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
81
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
82 for line in old_lines:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
83 if line is not None:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
84 print >> fh, line
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
85
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
86 fh.close()
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
87
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
88 if len(sys.argv) != 4:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
89 print >> sys.stderr, "Usage"
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
90 sys.exit(1)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
91
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
92 input, output, order_string = sys.argv[1:]
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
93
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
94 new_order = parse_rangelist(order_string)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
95 old_lines = get_lines(input)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
96 reorder(old_lines, new_order, output)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
97
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
98 sys.exit(0)