Mercurial > repos > miller-lab > genome_diversity

annotate specify.py @ 38:9d0b1fa77047

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Changed simplejson to json
author Richard Burhans <burhans@bx.psu.edu>
date Fri, 28 Feb 2014 12:15:32 -0500
parents 8997f2ca8c7a
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
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
3 import gd_util
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
4 import sys
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
5 from Population import Population
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
6
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
7 ################################################################################
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
8
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
9 def parse_string(str_arg, ind_token2col):
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
10 columns = []
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
11
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
12 string = gd_util.unwrap_string(str_arg)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
13 tokens = find_tokens(string, ind_token2col)
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
14
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
15 for token in tokens:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
16 col = ind_token2col[token]
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
17 if col not in columns:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
18 columns.append(col)
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
19
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
20 return columns
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
21
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
22 def find_tokens(string, tokens):
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
23 rv = []
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
24 for token in tokens:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
25 if token in string:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
26 if token not in rv:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
27 rv.append(token)
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
28 return rv
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
29
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
30 ################################################################################
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
31
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
32 if len(sys.argv) != 6:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
33 gd_util.die('Usage')
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
34
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
35 input, output, ind_arg, cb_arg, str_arg = sys.argv[1:]
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
36
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
37 p_total = Population()
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
38 p_total.from_wrapped_dict(ind_arg)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
39
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
40 p_cb = Population()
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
41 p_cb.from_wrapped_dict(cb_arg)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
42
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
43 if not p_total.is_superset(p_cb):
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
44 gd_util.die('There is a checked individual that does not appear in the SNP table')
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
45
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
46 ################################################################################
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
47
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
48 ind_col2name = {}
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
49 ind_token2col = {}
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
50 for col in p_total.column_list():
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
51 individual = p_total.individual_with_column(col)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
52 name = individual.name
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
53 ind_col2name[col] = name
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
54 first_token = name.split()[0]
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
55 if first_token not in ind_token2col:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
56 ind_token2col[first_token] = col
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
57 else:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
58 gd_util.die('duplicate first token: {0}'.format(first_token))
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
59
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
60 out_cols = p_cb.column_list()
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
61 str_cols = parse_string(str_arg, ind_token2col)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
62
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
63 with open(output, 'w') as fh:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
64 for col in sorted(ind_col2name.keys()):
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
65 if col in out_cols or col in str_cols:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 24
diff changeset
66 print >> fh, '\t'.join([str(x) for x in [col, ind_col2name[col], '']])
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
67
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
68 sys.exit(0)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
69