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