Mercurial > repos > devteam > sam_pileup
comparison sam_pileup.py @ 1:e7d863c5c5d6
Update sam_pileup to use the fasta_indexes data table.
author | Dave Bouvier <dave@bx.psu.edu> |
---|---|
date | Wed, 11 Dec 2013 12:54:32 -0500 |
parents | 95612c159681 |
children |
comparison
equal
deleted
inserted
replaced
0:95612c159681 | 1:e7d863c5c5d6 |
---|---|
6 usage: %prog [options] | 6 usage: %prog [options] |
7 -p, --input1=p: bam file | 7 -p, --input1=p: bam file |
8 -o, --output1=o: Output pileup | 8 -o, --output1=o: Output pileup |
9 -R, --ref=R: Reference file type | 9 -R, --ref=R: Reference file type |
10 -n, --ownFile=n: User-supplied fasta reference file | 10 -n, --ownFile=n: User-supplied fasta reference file |
11 -d, --dbkey=d: dbkey of user-supplied file | |
12 -x, --indexDir=x: Index directory | |
13 -b, --bamIndex=b: BAM index file | 11 -b, --bamIndex=b: BAM index file |
12 -g, --index=g: Path of the indexed reference genome | |
14 -s, --lastCol=s: Print the mapping quality as the last column | 13 -s, --lastCol=s: Print the mapping quality as the last column |
15 -i, --indels=i: Only output lines containing indels | 14 -i, --indels=i: Only output lines containing indels |
16 -M, --mapCap=M: Cap mapping quality | 15 -M, --mapCap=M: Cap mapping quality |
17 -c, --consensus=c: Call the consensus sequence using MAQ consensu model | 16 -c, --consensus=c: Call the consensus sequence using MAQ consensu model |
18 -T, --theta=T: Theta paramter (error dependency coefficient) | 17 -T, --theta=T: Theta paramter (error dependency coefficient) |
29 | 28 |
30 def stop_err( msg ): | 29 def stop_err( msg ): |
31 sys.stderr.write( '%s\n' % msg ) | 30 sys.stderr.write( '%s\n' % msg ) |
32 sys.exit() | 31 sys.exit() |
33 | 32 |
34 def check_seq_file( dbkey, GALAXY_DATA_INDEX_DIR ): | |
35 seqFile = '%s/sam_fa_indices.loc' % GALAXY_DATA_INDEX_DIR | |
36 seqPath = '' | |
37 for line in open( seqFile ): | |
38 line = line.rstrip( '\r\n' ) | |
39 if line and not line.startswith( '#' ) and line.startswith( 'index' ): | |
40 fields = line.split( '\t' ) | |
41 if len( fields ) < 3: | |
42 continue | |
43 if fields[1] == dbkey: | |
44 seqPath = fields[2].strip() | |
45 break | |
46 return seqPath | |
47 | |
48 def __main__(): | 33 def __main__(): |
49 #Parse Command Line | 34 #Parse Command Line |
50 options, args = doc_optparse.parse( __doc__ ) | 35 options, args = doc_optparse.parse( __doc__ ) |
51 seqPath = check_seq_file( options.dbkey, options.indexDir ) | |
52 # output version # of tool | 36 # output version # of tool |
53 try: | 37 try: |
54 tmp = tempfile.NamedTemporaryFile().name | 38 tmp = tempfile.NamedTemporaryFile().name |
55 tmp_stdout = open( tmp, 'wb' ) | 39 tmp_stdout = open( tmp, 'wb' ) |
56 proc = subprocess.Popen( args='samtools 2>&1', shell=True, stdout=tmp_stdout ) | 40 proc = subprocess.Popen( args='samtools 2>&1', shell=True, stdout=tmp_stdout ) |
75 tmpf0bam_name = '%s.bam' % tmpf0_name | 59 tmpf0bam_name = '%s.bam' % tmpf0_name |
76 tmpf0bambai_name = '%s.bam.bai' % tmpf0_name | 60 tmpf0bambai_name = '%s.bam.bai' % tmpf0_name |
77 tmpf1 = tempfile.NamedTemporaryFile( dir=tmpDir ) | 61 tmpf1 = tempfile.NamedTemporaryFile( dir=tmpDir ) |
78 tmpf1_name = tmpf1.name | 62 tmpf1_name = tmpf1.name |
79 tmpf1.close() | 63 tmpf1.close() |
80 tmpf1fai_name = '%s.fai' % tmpf1_name | |
81 #link bam and bam index to working directory (can't move because need to leave original) | 64 #link bam and bam index to working directory (can't move because need to leave original) |
82 os.symlink( options.input1, tmpf0bam_name ) | 65 os.symlink( options.input1, tmpf0bam_name ) |
83 os.symlink( options.bamIndex, tmpf0bambai_name ) | 66 os.symlink( options.bamIndex, tmpf0bambai_name ) |
84 #get parameters for pileup command | 67 #get parameters for pileup command |
85 if options.lastCol == 'yes': | 68 if options.lastCol == 'yes': |
98 try: | 81 try: |
99 # have to nest try-except in try-finally to handle 2.4 | 82 # have to nest try-except in try-finally to handle 2.4 |
100 try: | 83 try: |
101 #index reference if necessary and prepare pileup command | 84 #index reference if necessary and prepare pileup command |
102 if options.ref == 'indexed': | 85 if options.ref == 'indexed': |
103 if not os.path.exists( "%s.fai" % seqPath ): | 86 if not os.path.exists( "%s.fai" % options.index ): |
104 raise Exception, "No sequences are available for '%s', request them by reporting this error." % options.dbkey | 87 raise Exception, "Indexed genome %s not present, request it by reporting this error." % options.index |
105 cmd = cmd % ( opts, seqPath, tmpf0bam_name, options.output1 ) | 88 cmd = cmd % ( opts, options.index, tmpf0bam_name, options.output1 ) |
106 elif options.ref == 'history': | 89 elif options.ref == 'history': |
107 os.symlink( options.ownFile, tmpf1_name ) | 90 os.symlink( options.ownFile, tmpf1_name ) |
108 cmdIndex = 'samtools faidx %s' % ( tmpf1_name ) | 91 cmdIndex = 'samtools faidx %s' % ( tmpf1_name ) |
109 tmp = tempfile.NamedTemporaryFile( dir=tmpDir ).name | 92 tmp = tempfile.NamedTemporaryFile( dir=tmpDir ).name |
110 tmp_stderr = open( tmp, 'wb' ) | 93 tmp_stderr = open( tmp, 'wb' ) |