Mercurial > repos > aaronpetkau > flash

comparison assembly_stats_txt.py @ 0:44c401ebc424 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author aaronpetkau
date Sat, 04 Jul 2015 08:57:35 -0400
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:44c401ebc424
1 #!/usr/bin/env python
2 #Version 1.01 - bugs kindly corrected by Jan van Haarst
3 import pkg_resources
4 import logging, os, string, sys, tempfile, glob, shutil, types, urllib
5 import shlex, subprocess
6 from optparse import OptionParser, OptionGroup
7 from stat import *
8
9
10 log = logging.getLogger( __name__ )
11
12 assert sys.version_info[:2] >= ( 2, 4 )
13
14 def stop_err( msg ):
15 sys.stderr.write( "%s\n" % msg )
16 sys.exit()
17
18 def __main__():
19 #Parse Command Line
20 s = 'assembly_stats_txt.py: argv = %s\n' % (sys.argv)
21 argcnt = len(sys.argv)
22 html_file = sys.argv[1]
23 working_dir = sys.argv[2]
24 type = sys.argv[3]
25 bucket = sys.argv[4]
26 input = sys.argv[5]
27 stats = sys.argv[6]
28 sortedcontigs = sys.argv[7]
29 histogrampng = sys.argv[8]
30 summedcontigspng = sys.argv[9]
31 histogramdata = sys.argv[10]
32 summedcontigdata = sys.argv[11]
33 try: # for test - needs this done
34 os.makedirs(working_dir)
35 except Exception, e:
36 stop_err( 'Error running assembly_stats_txt.py ' + str( e ) )
37
38
39 cmdline = '%s/fasta_summary.pl -i %s -t %s %s -o %s > /dev/null' % (os.path.dirname(sys.argv[0]),input, type, bucket, working_dir)
40 try:
41 proc = subprocess.Popen( args=cmdline, shell=True, stderr=subprocess.PIPE )
42 returncode = proc.wait()
43 # get stderr, allowing for case where it's very large
44 stderr = ''
45 buffsize = 1048576
46 try:
47 while True:
48 stderr += proc.stderr.read( buffsize )
49 if not stderr or len( stderr ) % buffsize != 0:
50 break
51 except OverflowError:
52 pass
53 if returncode != 0:
54 raise Exception, stderr
55 except Exception, e:
56 stop_err( 'Error running assembly_stats.py ' + str( e ) )
57
58 stats_path = os.path.join(working_dir,'stats.txt')
59 sorted_contigs_path = os.path.join(working_dir,'sorted_contigs.fa')
60 histogram_png_path = os.path.join(working_dir,'histogram_bins.dat.png')
61 summed_contigs_path = os.path.join(working_dir,'summed_contig_lengths.dat.png')
62 histogram_data_path = os.path.join(working_dir,'histogram_bins.dat')
63 summed_contigs_data_path = os.path.join(working_dir,'summed_contig_lengths.dat')
64
65 out = open(stats,'w')
66 for line in open( stats_path ):
67 out.write( "%s" % (line) )
68 out.close()
69
70 out = open(sortedcontigs,'w')
71 for line in open(sorted_contigs_path ):
72 out.write( "%s" % (line) )
73 out.close()
74
75 out = open(histogrampng,'w')
76 for line in open(histogram_png_path ):
77 out.write( "%s" % (line) )
78 out.close()
79
80 out = open(summedcontigspng,'w')
81 for line in open(summed_contigs_path ):
82 out.write( "%s" % (line) )
83 out.close()
84
85
86 out = open(histogramdata,'w')
87 for line in open(histogram_data_path ):
88 out.write( "%s" % (line) )
89 out.close()
90
91 out = open(summedcontigdata,'w')
92 for line in open(summed_contigs_data_path ):
93 out.write( "%s" % (line) )
94 out.close()
95
96
97
98
99
100
101
102
103
104 # rval = ['<html><head><title>Assembly stats Galaxy Composite Dataset </title></head><p/>']
105 # rval.append('<div>%s<p/></div>' % (cmdline) )
106 # rval.append('<div>This composite dataset is composed of the following files:<p/><ul>')
107 # rval.append( '<li><a href="%s" type="text/plain">%s </a>%s</li>' % (stats_path,'stats.txt','stats.txt' ) )
108 # rval.append( '<li><a href="%s" type="text/plain">%s </a>%s</li>' % (sorted_contigs_path,'sorted_contigs.fa','sorted_contigs.fa' ) )
109 # rval.append( '<li><a href="%s" type="image/png">%s </a>%s</li>' % (histogram_png_path,'histogram_bins.dat.png','histogram_bins.dat.png' ) )
110 # rval.append( '<li><a href="%s" type="image/png">%s </a>%s</li>' % (summed_contigs_path,'summed_contig_lengths.dat.png','summed_contig_lengths.dat.png' ) )
111 # rval.append( '<li><a href="%s" type="text/plain">%s </a>%s</li>' % (histogram_data_path,'histogram_bins.dat','histogram_bins.dat' ) )
112 # rval.append( '<li><a href="%s" type="text/plain">%s </a>%s</li>' % (summed_contigs_data_path,'summed_contig_lengths.dat','summed_contig_lengths.dat' ) )
113
114
115 #
116 # rval.append( '</ul></div></html>' )
117 # f = file(html_file,'w')
118 # f.write("\n".join( rval ))
119 # f.write('\n')
120 # f.close()
121
122 if __name__ == "__main__": __main__()