Mercurial > repos > miller-lab > genome_diversity

annotate nucleotide_diversity_pi.py @ 24:248b06e86022

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Added gd_genotype datatype. Modified tools to support new datatype.
author Richard Burhans <burhans@bx.psu.edu>
date Tue, 28 May 2013 16:24:19 -0400
parents 95a05c1ef5d5
children 8997f2ca8c7a
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
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
3 import sys
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
4 import subprocess
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
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
9 def run_program(prog, args, stdout_file=None, space_to_tab=False):
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
10 #print "args: ", ' '.join(args)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
11 p = subprocess.Popen(args, bufsize=-1, executable=prog, stdin=None, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
12 (stdoutdata, stderrdata) = p.communicate()
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
13 rc = p.returncode
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
14
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
15 if stdout_file is not None:
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
16 with open(stdout_file, 'w') as ofh:
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
17 lines = stdoutdata.split('\n')
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
18 for line in lines:
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
19 line = line.strip()
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
20 if line:
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
21 if space_to_tab:
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
22 line = line.replace(' ', '\t')
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
23 print >> ofh, line
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 if rc != 0:
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
26 print >> sys.stderr, "FAILED: rc={0}: {1}".format(rc, ' '.join(args))
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
27 print >> sys.stderr, stderrdata
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
28 sys.exit(1)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
29
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
30 ################################################################################
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
31
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
32 if len(sys.argv) < 8:
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
33 print >> sys.stderr, "Usage"
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
34 sys.exit(1)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
35
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
36 gd_saps_file, gd_snps_file, covered_intervals_file, gd_indivs_file, output_file = sys.argv[1:6]
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
37 individual_metadata = sys.argv[6:]
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 p_total = Population()
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
40 p_total.from_tag_list(individual_metadata)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
41
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
42 p1 = Population()
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
43 p1.from_population_file(gd_indivs_file)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
44 if not p_total.is_superset(p1):
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
45 print >> sys.stderr, 'There is an individual in the population individuals that is not in the SNP table'
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
46 sys.exit(1)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
47
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
48 ################################################################################
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
49
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
50 prog = 'get_pi'
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
51
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
52 args = [ prog ]
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
53 args.append(gd_saps_file)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
54 args.append(gd_snps_file)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
55 args.append(covered_intervals_file)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
56
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
57 columns = p1.column_list()
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
58 for column in columns:
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
59 args.append('{0}'.format(column))
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
60
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
61 run_program(None, args, stdout_file=output_file)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
62
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
63 sys.exit(0)
95a05c1ef5d5 update to devshed revision aaece207bd01
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
64