Mercurial > repos > miller-lab > genome_diversity

comparison phylogenetic_tree.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 2c498d40ecde
children 8997f2ca8c7a
comparison
equal deleted inserted replaced
23:66a183c44dd5 24:248b06e86022
17 if e.errno <> errno.EEXIST: 17 if e.errno <> errno.EEXIST:
18 raise 18 raise
19 19
20 ################################################################################ 20 ################################################################################
21 21
22 if len(sys.argv) < 11: 22 # <command interpreter="python">
23 print >> sys.stderr, "Usage" 23 # phylogenetic_tree.py "$input" "$output" "$output.files_path"
24 #
25 # #if $input_type.choice == '0'
26 # "gd_snp"
27 # #if $input_type.data_source.choice == '0'
28 # "sequence_coverage"
29 # "$input_type.data_source.minimum_coverage"
30 # "$input_type.data_source.minimum_quality"
31 # #else if $input_type.data_source.choice == '1'
32 # "estimated_genotype"
33 # #else if $input_type.choice == '1'
34 # "gd_genotype"
35 # #end if
36 #
37 # #if $individuals.choice == '0'
38 # "all_individuals"
39 # #else if $individuals.choice == '1'
40 # "$individuals.p1_input"
41 # #end if
42 #
43 # #if ((str($input.metadata.scaffold) == str($input.metadata.ref)) and (str($input.metadata.pos) == str($input.metadata.rPos))) or (str($include_reference) == '0')
44 # "none"
45 # #else
46 # "$input.metadata.dbkey"
47 # #end if
48 #
49 # #set $draw_tree_options = ''.join(str(x) for x in [$branch_style, $scale_style, $length_style, $layout_style])
50 # #if $draw_tree_options == ''
51 # ""
52 # #else
53 # "-$draw_tree_options"
54 # #end if
55 #
56 # #for $individual_name, $individual_col in zip($input.dataset.metadata.individual_names, $input.dataset.metadata.individual_columns)
57 # #set $arg = '%s:%s' % ($individual_col, $individual_name)
58 # "$arg"
59 # #end for
60 # </command>
61
62 ################################################################################
63
64 # if len(sys.argv) < 11:
65 # print >> sys.stderr, "Usage"
66 # sys.exit(1)
67 #
68 # input, p1_input, output, extra_files_path, minimum_coverage, minimum_quality, dbkey, data_source, draw_tree_options = sys.argv[1:10]
69 #
70 # individual_metadata = sys.argv[10:]
71 #
72 # # note: TEST THIS
73 # if dbkey in ['', '?', 'None']:
74 # dbkey = 'none'
75 #
76 # p_total = Population()
77 # p_total.from_tag_list(individual_metadata)
78
79 if len(sys.argv) < 5:
80 print >> sys.stderr, 'Usage'
24 sys.exit(1) 81 sys.exit(1)
25 82
26 input, p1_input, output, extra_files_path, minimum_coverage, minimum_quality, dbkey, data_source, draw_tree_options = sys.argv[1:10] 83 input, output, extra_files_path, input_type = sys.argv[1:5]
27 84 args = sys.argv[5:]
28 individual_metadata = sys.argv[10:] 85
86 data_source = '1'
87 minimum_coverage = '0'
88 minimum_quality = '0'
89
90 if input_type == 'gd_snp':
91 data_source_arg = args.pop(0)
92 if data_source_arg == 'sequence_coverage':
93 data_source = '0'
94 minimum_coverage = args.pop(0)
95 minimum_quality = args.pop(0)
96 elif data_source_arg == 'estimated_genotype':
97 pass
98 else:
99 print >> sys.stderr, 'Unsupported data_source:', data_source_arg
100 sys.exit(1)
101 elif input_type == 'gd_genotype':
102 pass
103 else:
104 print >> sys.stderr, 'Unsupported input_type:', input_type
105 sys.exit(1)
106
107 p1_input, dbkey, draw_tree_options = args[:3]
29 108
30 # note: TEST THIS 109 # note: TEST THIS
31 if dbkey in ['', '?', 'None']: 110 if dbkey in ['', '?', 'None']:
32 dbkey = 'none' 111 dbkey = 'none'
33 112
113 individual_metadata = args[3:]
114
34 p_total = Population() 115 p_total = Population()
35 p_total.from_tag_list(individual_metadata) 116 p_total.from_tag_list(individual_metadata)
36
37 117
38 ################################################################################ 118 ################################################################################
39 119
40 mkdir_p(extra_files_path) 120 mkdir_p(extra_files_path)
41 121
86 print >> sys.stderr, 'There is an individual in the population that is not in the SNP table' 166 print >> sys.stderr, 'There is an individual in the population that is not in the SNP table'
87 sys.exit(1) 167 sys.exit(1)
88 tags = p1.tag_list() 168 tags = p1.tag_list()
89 169
90 for tag in tags: 170 for tag in tags:
171 if input_type == 'gd_genotype':
172 column, name = tag.split(':')
173 tag = '{0}:{1}'.format(int(column) - 2, name)
91 args.append(tag) 174 args.append(tag)
92 175
93 fh = open(phylip_outfile, 'w') 176 fh = open(phylip_outfile, 'w')
94 run_program(None, args, fh) 177 run_program(None, args, fh)
95 178