Mercurial > repos > jjohnson > barcode_splitter
annotate fastx_barcode_splitter_single_galaxy_wrapper.py @ 0:bc23f6946bb8 default tip
Alternative barcode splitters that move selected results to the users history.
author | Jim Johnson <jj@umn.edu> |
---|---|
date | Tue, 19 Jul 2011 13:03:32 -0500 |
parents | |
children |
rev | line source |
---|---|
0
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
1 import sys, os, os.path, tempfile, shutil, re, shlex, subprocess |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
2 |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
3 def stop_err( msg ): |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
4 sys.stderr.write( "%s\n" % msg ) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
5 sys.exit() |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
6 |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
7 # tranform fastx_barcode_splitter result to html |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
8 def results_to_html(results_path,html_path,basepath,print_stdout ): |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
9 pat = '%s[/]?([^\t]*)' % basepath |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
10 rep = '<a href=\"\\1\">\\1</a>' |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
11 txt = open(results_path,'r') |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
12 html = open(html_path,'w') |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
13 html.write('<html><body><table border=1>\n') |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
14 try: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
15 for line in txt: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
16 html.write('<tr><td>%s</td></tr>' % re.sub('\t','</td><td>',re.sub(pat,rep,line))) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
17 if print_stdout: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
18 print >> sys.stdout, '\t'.join(line.split('\t')[:2]) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
19 except Exception, e: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
20 print(str(e)) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
21 pass |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
22 html.write('</table></body></html>\n') |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
23 html.close() |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
24 txt.close() |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
25 |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
26 def __main__(): |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
27 """ |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
28 ##params for galaxy wrapper |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
29 $output |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
30 "$output.id" |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
31 "$input.ext" |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
32 "$__new_file_path__" |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
33 --barcodes='$barcodes' |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
34 $BARCODE $input "$input.name" "$output.extra_files_path" |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
35 ## params for fastx_barcode_splitter |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
36 --mismatches $mismatches --partial $partial $EOL |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
37 """ |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
38 |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
39 output = sys.argv[1] |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
40 output_unmatched = sys.argv[2] |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
41 file_ext = sys.argv[3] |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
42 select_barcode = sys.argv[4].replace('--barcodes=','') |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
43 barcodes = os.path.abspath("barcodes") |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
44 with open(barcodes, 'w') as f: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
45 f.write("barcode\t%s\n" % (select_barcode)) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
46 |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
47 #barcodes = sys.argv[6] |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
48 fastx = sys.argv[5] |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
49 fastx_name = sys.argv[6] |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
50 #extra_files_path = sys.argv[9] |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
51 script_args = ' '.join(sys.argv[7:]) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
52 #Sanitize library name, make sure we can create a file with this name |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
53 lib_name = re.sub('\W','_',re.sub('\.\W*$','',fastx_name))+'_' |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
54 # Check that input datasets exist |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
55 if not os.path.isfile(fastx): |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
56 stop_err('Error: Input file (%s) not found!' % fastx) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
57 try: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
58 prefix = lib_name |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
59 cmd_line = 'zcat -f %s | fastx_barcode_splitter.pl --bcfile %s --prefix %s --suffix %s %s' %(fastx,barcodes,prefix,'.'+file_ext,script_args) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
60 # print >> sys.stderr, cmd_line |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
61 # Create file to collect results written to stdout |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
62 tmp_dir = tempfile.mkdtemp() |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
63 result_path = tempfile.NamedTemporaryFile(dir=tmp_dir, prefix='results_', suffix='.out' ).name |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
64 result_file = open( result_path, 'wb' ) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
65 proc = subprocess.Popen( args=cmd_line, shell=True, cwd=tmp_dir, stderr=subprocess.PIPE,stdout=result_file.fileno() ) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
66 returncode = proc.wait() |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
67 result_file.close() |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
68 stderr = proc.stderr.read() |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
69 if returncode != 0: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
70 raise Exception, stderr |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
71 # copy results to ouptut |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
72 #results_to_html(result_path,output,extra_files_path,True) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
73 # make new datasets for selected barcodes |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
74 flist = os.listdir(tmp_dir) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
75 for fname in flist: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
76 if fname.find('_'+barcode+'.'+file_ext) >= 0: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
77 fpath = os.path.join(tmp_dir,fname) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
78 shutil.copy2(fpath, output) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
79 for fname in flist: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
80 if fname.find('_unmatched.' + file_ext) > 0: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
81 fpath = os.path.join(tmp_dir, fname) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
82 shutil.copy2(fpath, output_unmatched) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
83 except Exception, e: |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
84 raise Exception, 'Exception caught attempting conversion: ' + str( e ) |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
85 |
bc23f6946bb8
Alternative barcode splitters that move selected results to the users history.
Jim Johnson <jj@umn.edu>
parents:
diff
changeset
|
86 if __name__ == "__main__": __main__() |