0
|
1 #!/usr/bin/env python
|
|
2 # Dave B.
|
|
3 # Uses fasta sorting functions written by Dan Blankenberg.
|
|
4
|
|
5 import json
|
|
6 import optparse
|
|
7 import os
|
|
8 import shutil
|
|
9 import subprocess
|
|
10 import sys
|
|
11 import tempfile
|
|
12
|
|
13 CHUNK_SIZE = 2**20
|
|
14
|
|
15 DEFAULT_DATA_TABLE_NAME = "fasta_indexes"
|
|
16
|
|
17 def get_id_name( params, dbkey, fasta_description=None):
|
|
18 #TODO: ensure sequence_id is unique and does not already appear in location file
|
|
19 sequence_id = params['param_dict']['sequence_id']
|
|
20 if not sequence_id:
|
|
21 sequence_id = dbkey
|
|
22
|
|
23 sequence_name = params['param_dict']['sequence_name']
|
|
24 if not sequence_name:
|
|
25 sequence_name = fasta_description
|
|
26 if not sequence_name:
|
|
27 sequence_name = dbkey
|
|
28 return sequence_id, sequence_name
|
|
29
|
|
30 def build_picard_index( data_manager_dict, fasta_filename, target_directory, dbkey, sequence_id, sequence_name, jar, data_table_name=DEFAULT_DATA_TABLE_NAME ):
|
|
31 fasta_base_name = os.path.split( fasta_filename )[-1]
|
|
32 gatk_sorted_fasta_filename = os.path.join( target_directory, fasta_base_name )
|
|
33 shutil.copy( fasta_filename, gatk_sorted_fasta_filename )
|
|
34 _sort_fasta_gatk( gatk_sorted_fasta_filename )
|
|
35 sam_index_filename = '%s.fai' % gatk_sorted_fasta_filename
|
|
36 if not os.path.exists( sam_index_filename ):
|
|
37 sam_command = [ 'samtools', 'faidx', gatk_sorted_fasta_filename ]
|
|
38 _run_command( sam_command, target_directory )
|
|
39 args = [ 'java', '-jar', jar, 'R=%s' % gatk_sorted_fasta_filename, 'O=%s.dict' % sequence_id ]
|
|
40 _run_command( args, target_directory )
|
|
41 data_table_entry = dict( value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name )
|
|
42 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
|
|
43
|
|
44 def _run_command( command, target_directory ):
|
|
45 tmp_stderr = tempfile.NamedTemporaryFile( prefix = "tmp-data-manager-gatk_picard_index_builder-stderr" )
|
|
46 proc = subprocess.Popen( args=command, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno() )
|
|
47 return_code = proc.wait()
|
|
48 if return_code:
|
|
49 tmp_stderr.flush()
|
|
50 tmp_stderr.seek( 0 )
|
|
51 sys.stderr.write( "Error building index:\n" )
|
|
52 while True:
|
|
53 chunk = tmp_stderr.read( CHUNK_SIZE )
|
|
54 if not chunk:
|
|
55 break
|
|
56 sys.stderr.write( chunk )
|
|
57 sys.exit( return_code )
|
|
58 tmp_stderr.close()
|
|
59
|
|
60 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
|
|
61 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
|
|
62 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
|
|
63 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
|
|
64 return data_manager_dict
|
|
65
|
|
66 def _move_and_index_fasta_for_sorting( fasta_filename ):
|
|
67 unsorted_filename = tempfile.NamedTemporaryFile().name
|
|
68 shutil.move( fasta_filename, unsorted_filename )
|
|
69 fasta_offsets = {}
|
|
70 unsorted_fh = open( unsorted_filename )
|
|
71 while True:
|
|
72 offset = unsorted_fh.tell()
|
|
73 line = unsorted_fh.readline()
|
|
74 if not line:
|
|
75 break
|
|
76 if line.startswith( ">" ):
|
|
77 line = line.split( None, 1 )[0][1:]
|
|
78 fasta_offsets[ line ] = offset
|
|
79 unsorted_fh.close()
|
|
80 current_order = map( lambda x: x[1], sorted( map( lambda x: ( x[1], x[0] ), fasta_offsets.items() ) ) )
|
|
81 return ( unsorted_filename, fasta_offsets, current_order )
|
|
82
|
|
83 def _write_sorted_fasta( sorted_names, fasta_offsets, sorted_fasta_filename, unsorted_fasta_filename ):
|
|
84 unsorted_fh = open( unsorted_fasta_filename )
|
|
85 sorted_fh = open( sorted_fasta_filename, 'wb+' )
|
|
86
|
|
87 for name in sorted_names:
|
|
88 offset = fasta_offsets[ name ]
|
|
89 unsorted_fh.seek( offset )
|
|
90 sorted_fh.write( unsorted_fh.readline() )
|
|
91 while True:
|
|
92 line = unsorted_fh.readline()
|
|
93 if not line or line.startswith( ">" ):
|
|
94 break
|
|
95 sorted_fh.write( line )
|
|
96 unsorted_fh.close()
|
|
97 sorted_fh.close()
|
|
98
|
|
99 def _int_to_roman( integer ):
|
|
100 if not isinstance( integer, int ):
|
|
101 raise TypeError, "expected integer, got %s" % type( integer )
|
|
102 if not 0 < integer < 4000:
|
|
103 raise ValueError, "Argument must be between 1 and 3999, got %s" % str( integer )
|
|
104 ints = ( 1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1 )
|
|
105 nums = ( 'M', 'CM', 'D', 'CD', 'C', 'XC', 'L', 'XL', 'X', 'IX', 'V', 'IV', 'I' )
|
|
106 result = ""
|
|
107 for i in range( len( ints ) ):
|
|
108 count = int( integer / ints[ i ] )
|
|
109 result += nums[ i ] * count
|
|
110 integer -= ints[ i ] * count
|
|
111 return result
|
|
112
|
|
113 def _sort_fasta_gatk( fasta_filename ):
|
|
114 ( unsorted_filename, fasta_offsets, current_order ) = _move_and_index_fasta_for_sorting( fasta_filename )
|
|
115 sorted_names = map( str, range( 1, 100 ) ) + map( _int_to_roman, range( 1, 100 ) ) + [ 'X', 'Y', 'M' ]
|
|
116 #detect if we have chrN, or just N
|
|
117 has_chr = False
|
|
118 for chrom in sorted_names:
|
|
119 if "chr%s" % chrom in current_order:
|
|
120 has_chr = True
|
|
121 break
|
|
122
|
|
123 if has_chr:
|
|
124 sorted_names = map( lambda x: "chr%s" % x, sorted_names)
|
|
125 else:
|
|
126 sorted_names.insert( 0, "MT" )
|
|
127 sorted_names.extend( map( lambda x: "%s_random" % x, sorted_names ) )
|
|
128
|
|
129 existing_sorted_names = []
|
|
130 for name in sorted_names:
|
|
131 # Append each chromosome only once.
|
|
132 if name in current_order and name not in existing_sorted_names:
|
|
133 existing_sorted_names.append( name )
|
|
134 for name in current_order:
|
|
135 #TODO: confirm that non-canonical names do not need to be sorted specially
|
|
136 if name not in existing_sorted_names:
|
|
137 existing_sorted_names.append( name )
|
|
138
|
|
139 if existing_sorted_names == current_order:
|
|
140 shutil.move( unsorted_filename, fasta_filename )
|
|
141 else:
|
|
142 _write_sorted_fasta( existing_sorted_names, fasta_offsets, fasta_filename, unsorted_filename )
|
|
143
|
|
144 def main():
|
|
145 #Parse Command Line
|
|
146 parser = optparse.OptionParser()
|
|
147 parser.add_option( '-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename' )
|
|
148 parser.add_option( '-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey' )
|
|
149 parser.add_option( '-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description' )
|
|
150 parser.add_option( '-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name' )
|
|
151 parser.add_option( '-j', '--jar', dest='jar', action='store', type="string", default=None, help='GATK .jar file' )
|
|
152 (options, args) = parser.parse_args()
|
|
153
|
|
154 filename = args[0]
|
|
155
|
|
156 params = json.loads( open( filename ).read() )
|
|
157 target_directory = params[ 'output_data' ][0]['extra_files_path']
|
|
158 os.mkdir( target_directory )
|
|
159 data_manager_dict = {}
|
|
160
|
|
161 if options.fasta_dbkey in [ None, '', '?' ]:
|
|
162 raise Exception( '"%s" is not a valid dbkey. You must specify a valid dbkey.' % ( options.fasta_dbkey ) )
|
|
163
|
|
164 sequence_id, sequence_name = get_id_name( params, dbkey=options.fasta_dbkey, fasta_description=options.fasta_description )
|
|
165
|
|
166 #build the index
|
|
167 build_picard_index( data_manager_dict,
|
|
168 options.fasta_filename,
|
|
169 target_directory,
|
|
170 options.fasta_dbkey,
|
|
171 sequence_id,
|
|
172 sequence_name,
|
|
173 options.jar,
|
|
174 data_table_name=options.data_table_name or DEFAULT_DATA_TABLE_NAME )
|
|
175
|
|
176 #save info to json file
|
|
177 open( filename, 'wb' ).write( json.dumps( data_manager_dict ) )
|
|
178
|
|
179 if __name__ == "__main__": main()
|