Mercurial > repos > devteam > pileup_interval

annotate pileup_interval.py @ 3:d78f28cae91b draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
author devteam
date Fri, 15 Jan 2021 11:38:35 +0000
parents 8afc93a5f9ae
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
1 #!/usr/bin/env python
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
2
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
3 """
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
4 Condenses pileup format into ranges of bases.
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
5
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
6 usage: %prog [options]
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
7 -i, --input=i: Input pileup file
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
8 -o, --output=o: Output pileup
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
9 -c, --coverage=c: Coverage
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
10 -f, --format=f: Pileup format
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
11 -b, --base=b: Base to select
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
12 -s, --seq_column=s: Sequence column
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
13 -l, --loc_column=l: Base location column
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
14 -r, --base_column=r: Reference base column
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
15 -C, --cvrg_column=C: Coverage column
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
16 """
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
17 import sys
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
18
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
19 from bx.cookbook import doc_optparse
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
20
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
21
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
22 def __main__():
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
23 # Parse Command Line
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
24 options, args = doc_optparse.parse(__doc__)
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
25 coverage = int(options.coverage)
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
26 fin = open(options.input, 'r')
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
27 fout = open(options.output, 'w')
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
28 inLine = fin.readline()
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
29 if options.format == 'six':
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
30 seqIndex = 0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
31 locIndex = 1
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
32 baseIndex = 2
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
33 covIndex = 3
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
34 elif options.format == 'ten':
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
35 seqIndex = 0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
36 locIndex = 1
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
37 if options.base == 'first':
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
38 baseIndex = 2
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
39 else:
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
40 baseIndex = 3
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
41 covIndex = 7
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
42 else:
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
43 seqIndex = int(options.seq_column) - 1
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
44 locIndex = int(options.loc_column) - 1
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
45 baseIndex = int(options.base_column) - 1
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
46 covIndex = int(options.cvrg_column) - 1
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
47 lastSeq = ''
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
48 lastLoc = -1
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
49 locs = []
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
50 startLoc = -1
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
51 bases = []
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
52 while inLine.strip() != '':
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
53 lineParts = inLine.split('\t')
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
54 try:
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
55 seq, loc, base, cov = lineParts[seqIndex], int(lineParts[locIndex]), lineParts[baseIndex], int(lineParts[covIndex])
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
56 except IndexError as ei:
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
57 if options.format == 'ten':
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
58 sys.exit('It appears that you have selected 10 columns while your file has 6. Make sure that the number of columns you specify matches the number in your file.\n' + str(ei))
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
59 else:
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
60 sys.exit('There appears to be something wrong with your column index values.\n' + str(ei))
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
61 except ValueError as ev:
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
62 if options.format == 'six':
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
63 sys.exit('It appears that you have selected 6 columns while your file has 10. Make sure that the number of columns you specify matches the number in your file.\n' + str(ev))
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
64 else:
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
65 sys.exit('There appears to be something wrong with your column index values.\n' + str(ev))
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
66 # strout += str(startLoc) + '\n'
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
67 # strout += str(bases) + '\n'
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
68 # strout += '%s\t%s\t%s\t%s\n' % (seq, loc, base, cov)
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
69 if loc == lastLoc + 1 or lastLoc == -1:
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
70 if cov >= coverage:
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
71 if seq == lastSeq or lastSeq == '':
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
72 if startLoc == -1:
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
73 startLoc = loc
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
74 locs.append(loc)
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
75 bases.append(base)
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
76 else:
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
77 if len(bases) > 0:
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
78 fout.write('%s\t%s\t%s\t%s\n' % (lastSeq, startLoc - 1, lastLoc, ''.join(bases)))
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
79 startLoc = loc
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
80 locs = [loc]
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
81 bases = [base]
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
82 else:
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
83 if len(bases) > 0:
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
84 fout.write('%s\t%s\t%s\t%s\n' % (lastSeq, startLoc - 1, lastLoc, ''.join(bases)))
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
85 startLoc = -1
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
86 locs = []
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
87 bases = []
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
88 else:
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
89 if len(bases) > 0:
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
90 fout.write('%s\t%s\t%s\t%s\n' % (lastSeq, startLoc - 1, lastLoc, ''.join(bases)))
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
91 if cov >= coverage:
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
92 startLoc = loc
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
93 locs = [loc]
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
94 bases = [base]
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
95 else:
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
96 startLoc = -1
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
97 locs = []
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
98 bases = []
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
99 lastSeq = seq
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
100 lastLoc = loc
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
101 inLine = fin.readline()
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
102 if len(bases) > 0:
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
103 fout.write('%s\t%s\t%s\t%s\n' % (lastSeq, startLoc - 1, lastLoc, ''.join(bases)))
0
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
104 fout.close()
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
105 fin.close()
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
106 # import sys
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
107 # strout += file(fout.name,'r').read()
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
108 # sys.stderr.write(strout)
8afc93a5f9ae Uploaded tool tarball.
devteam
parents:
diff changeset
109
3
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
110
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
111 if __name__ == "__main__":
d78f28cae91b "planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/pileup_interval commit 8b2095c59ecc2e94c58a42e2e04dbcecdc823dbf"
devteam
parents: 0
diff changeset
112 __main__()