Mercurial > repos > miller-lab > genome_diversity

annotate aggregate_gd_indivs.py @ 32:03c22b722882

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
remove BeautifulSoup dependency
author Richard Burhans <burhans@bx.psu.edu>
date Fri, 20 Sep 2013 13:54:23 -0400
parents 8997f2ca8c7a
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
22
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
1 #!/usr/bin/env python
95a05c1ef5d5 update to devshed revision aaece207bd01
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: 26
diff changeset
3 import gd_util
22
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
4 import sys
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
5 from Population import Population
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
6
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
7 ################################################################################
95a05c1ef5d5 update to devshed revision aaece207bd01
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: 26
diff changeset
9 if len(sys.argv) != 6:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 26
diff changeset
10 gd_util.dir('Usage')
22
95a05c1ef5d5 update to devshed revision aaece207bd01
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: 26
diff changeset
12 input, p1_input, output, input_type, ind_arg = sys.argv[1:]
22
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
13
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
14 p_total = Population()
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 26
diff changeset
15 p_total.from_wrapped_dict(ind_arg)
22
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
16
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
17 p1 = Population()
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
18 p1.from_population_file(p1_input)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
19
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
20 if not p_total.is_superset(p1):
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 26
diff changeset
21 gd_util.die('There is an individual in the population that is not in the SNP table')
22
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
22
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
23 ################################################################################
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
24
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
25 prog = 'aggregate'
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
26
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 26
diff changeset
27 args = [ prog ]
22
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
28 args.append(input)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
29
26
91e835060ad2 Updates to Admixture, Aggregate Individuals, and Restore Attributes to support gd_genotype
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
30 if input_type == 'gd_snp':
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 26
diff changeset
31 args.append(1)
26
91e835060ad2 Updates to Admixture, Aggregate Individuals, and Restore Attributes to support gd_genotype
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
32 elif input_type == 'gd_genotype':
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 26
diff changeset
33 args.append(0)
26
91e835060ad2 Updates to Admixture, Aggregate Individuals, and Restore Attributes to support gd_genotype
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
34 else:
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 26
diff changeset
35 die('unknown input type: {0}'.format(input_type))
26
91e835060ad2 Updates to Admixture, Aggregate Individuals, and Restore Attributes to support gd_genotype
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
36
22
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
37 columns = p1.column_list()
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
38
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
39 for column in sorted(columns):
26
91e835060ad2 Updates to Admixture, Aggregate Individuals, and Restore Attributes to support gd_genotype
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
40 if input_type == 'gd_genotype':
91e835060ad2 Updates to Admixture, Aggregate Individuals, and Restore Attributes to support gd_genotype
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
41 column = str(int(column) - 2)
22
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
42 args.append(column)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
43
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 26
diff changeset
44 with open(output, 'w') as fh:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents: 26
diff changeset
45 gd_util.run_program(prog, args, stdout=fh)
22
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
46
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
47 sys.exit(0)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
48