annotate utilities/validateBam.py @ 10:7d10b55965c9 draft default tip

planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
author thondeboer
date Wed, 16 May 2018 17:02:51 -0400
parents 6e75a84e9338
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
1 #!/usr/bin/env python
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
2
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
3 import sys
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
4 import os
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
5 import gzip
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
6 from struct import unpack
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
7
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
8 BAM_EOF = ['1f', '8b', '08', '04', '00', '00', '00', '00', '00', 'ff', '06', '00', '42', '43', '02', '00', '1b', '00', '03', '00', '00', '00', '00', '00', '00', '00', '00', '00']
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
9
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
10 def getBytes(fmt,amt):
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
11 if fmt == '<i' or fmt == '<I':
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
12 mySize = 4
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
13 elif fmt == '<c' or fmt == '<b' or fmt == '<B':
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
14 mySize = 1
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
15 else:
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
16 print '\nError, unknown format:',fmt,'\n'
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
17 exit(1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
18 if amt == 1:
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
19 fread = f.read(mySize)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
20 if not fread:
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
21 return None
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
22 return unpack(fmt,fread)[0]
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
23 else:
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
24 fread = f.read(mySize*amt)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
25 if not fread:
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
26 return None
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
27 return unpack(fmt,fread)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
28
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
29 # check eof
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
30 IN_BAM = sys.argv[1]
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
31 f = open(IN_BAM,'rb')
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
32 f.seek(os.path.getsize(IN_BAM)-28)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
33 EOF = [format(ord(n),'02x') for n in f.read()]
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
34 print 'EOF_MARKER: ', ' '.join(EOF)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
35 if EOF != BAM_EOF:
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
36 print '\nWARNING: BAM EOF DOES NOT MATCH EXPECTED STRING.\n'
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
37 f.close()
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
38
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
39 # check other stuff
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
40 f = gzip.open(IN_BAM,'rb')
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
41
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
42 print 'MAGIC STRING:', f.read(4)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
43 l_text = getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
44 print 'l_text: ', l_text
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
45 print 'text: \n', f.read(l_text)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
46 n_ref = getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
47 print 'n_ref: ', n_ref
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
48
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
49 for i in xrange(n_ref):
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
50 l_name = getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
51 print 'ref'+str(i)+' - l_name:', l_name
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
52 print 'ref'+str(i)+' - name: ', f.read(l_name)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
53 print 'ref'+str(i)+' - l_ref: ', getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
54
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
55 print '\nEXAMINING ALIGNMENT DATA:\n'
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
56 aln_N = 0
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
57 while True:
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
58 aln_N += 1
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
59 blockSize = getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
60 if blockSize == None:
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
61 break
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
62 print '['+str(aln_N)+']:', 'blockSize:', blockSize
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
63 print '-- refID:', getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
64 print '-- pos: ', getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
65 bmqnl = getBytes('<I',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
66 binv = (bmqnl>>16)&65535
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
67 mapq = (bmqnl>>8)&255
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
68 lrn = bmqnl&255
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
69 print '-- bmqnl:', bmqnl, '(bin='+str(binv)+', mapq='+str(mapq)+', l_readname+1='+str(lrn)+')'
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
70 flgnc = getBytes('<I',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
71 flag = (flgnc>>16)&65535
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
72 ncig = flgnc&65535
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
73 print '-- flgnc:', flgnc, '(flag='+str(flag)+', ncig='+str(ncig)+')'
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
74 print '-- l_seq:', getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
75 print '-- nxtID:', getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
76 print '-- nxtPo:', getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
77 print '-- tlen: ', getBytes('<i',1)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
78 print '-- rname:', str([f.read(lrn)])[1:-1]
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
79
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
80 f.read(blockSize-32-lrn)
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
81 #print [block]
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
82
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
83
6e75a84e9338 planemo upload commit e96b43f96afce6a7b7dfd4499933aad7d05c955e-dirty
thondeboer
parents:
diff changeset
84 f.close()