annotate extract_genomic_dna_utils.py @ 10:5cc8e93ee98f draft default tip

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit a656cf7893ea9ce63d0e470c027fc185b5b6d82d"
author iuc
date Fri, 06 Dec 2019 15:24:00 -0500
parents 3088e7e70888
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
1 import copy
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
2 import os
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
3 import subprocess
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
4 import sys
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
5 import tempfile
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
6
3
b71579ad576c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit bd4fbe120288bf8452e479cbd82aa1bbf5c4bd31
iuc
parents: 2
diff changeset
7 from bx.intervals.io import Comment, GenomicInterval, Header
0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
8 from bx.intervals.io import GenomicIntervalReader, NiceReaderWrapper, ParseError
7
3088e7e70888 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 3
diff changeset
9 from six import Iterator
0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
10
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
11 # Default chrom, start, end, strand cols for a bed file
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
12 BED_DEFAULT_COLS = 0, 1, 2, 5
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
13
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
14
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
15 class GFFInterval(GenomicInterval):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
16 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
17 A GFF interval, including attributes. If file is strictly a GFF file,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
18 only attribute is 'group.'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
19 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
20
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
21 def __init__(self, reader, fields, chrom_col=0, feature_col=2, start_col=3, end_col=4,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
22 strand_col=6, score_col=5, default_strand='.', fix_strand=False):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
23 # GFF format allows '.' for strand but GenomicInterval does not. To get around this,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
24 # temporarily set strand and then unset after initing GenomicInterval.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
25 unknown_strand = False
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
26 if not fix_strand and fields[strand_col] == '.':
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
27 unknown_strand = True
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
28 fields[strand_col] = '+'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
29 GenomicInterval.__init__(self, reader, fields, chrom_col, start_col, end_col,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
30 strand_col, default_strand, fix_strand=fix_strand)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
31 if unknown_strand:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
32 self.strand = '.'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
33 self.fields[strand_col] = '.'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
34 # Handle feature, score column.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
35 self.feature_col = feature_col
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
36 if self.feature_col >= self.nfields:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
37 stop_err("No field for feature_col (%d)" % feature_col)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
38 self.feature = self.fields[self.feature_col]
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
39 self.score_col = score_col
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
40 if self.score_col >= self.nfields:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
41 stop_err("No field for score_col (%d)" % score_col)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
42 self.score = self.fields[self.score_col]
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
43 # GFF attributes.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
44 self.attributes = parse_gff_attributes(fields[8])
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
45
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
46 def copy(self):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
47 return GFFInterval(self.reader, list(self.fields), self.chrom_col, self.feature_col,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
48 self.start_col, self.end_col, self.strand_col, self.score_col, self.strand)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
49
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
50
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
51 class GFFFeature(GFFInterval):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
52 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
53 A GFF feature, which can include multiple intervals.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
54 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
55
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
56 def __init__(self, reader, chrom_col=0, feature_col=2, start_col=3, end_col=4, strand_col=6,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
57 score_col=5, default_strand='.', fix_strand=False, intervals=[], raw_size=0):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
58 # Use copy so that first interval and feature do not share fields.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
59 GFFInterval.__init__(self, reader, copy.deepcopy(intervals[0].fields), chrom_col, feature_col,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
60 start_col, end_col, strand_col, score_col, default_strand, fix_strand=fix_strand)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
61 self.intervals = intervals
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
62 self.raw_size = raw_size
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
63 # Use intervals to set feature attributes.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
64 for interval in self.intervals:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
65 # Error checking. NOTE: intervals need not share the same strand.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
66 if interval.chrom != self.chrom:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
67 stop_err("interval chrom does not match self chrom: %s != %s" % (interval.chrom, self.chrom))
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
68 # Set start, end of interval.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
69 if interval.start < self.start:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
70 self.start = interval.start
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
71 if interval.end > self.end:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
72 self.end = interval.end
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
73
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
74 def name(self):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
75 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
76 Returns feature's name.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
77 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
78 name = None
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
79 # Preference for name:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
80 # GTF: 'gene_id', 'transcript_id'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
81 # GFF3: 'ID', 'id'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
82 # GFF: 'group'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
83 for attr_name in ['gene_id', 'transcript_id', 'ID', 'id', 'group']:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
84 name = self.attributes.get(attr_name, None)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
85 if name is not None:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
86 break
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
87 return name
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
88
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
89 def copy(self):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
90 intervals_copy = []
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
91 for interval in self.intervals:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
92 intervals_copy.append(interval.copy())
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
93 return GFFFeature(self.reader, self.chrom_col, self.feature_col, self.start_col, self.end_col,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
94 self.strand_col, self.score_col, self.strand, intervals=intervals_copy)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
95
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
96 def lines(self):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
97 lines = []
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
98 for interval in self.intervals:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
99 lines.append('\t'.join(interval.fields))
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
100 return lines
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
101
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
102
7
3088e7e70888 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 3
diff changeset
103 class GFFReaderWrapper(Iterator, NiceReaderWrapper):
0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
104 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
105 Reader wrapper for GFF files which has two major functions:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
106 1. group entries for GFF file (via group column), GFF3 (via id attribute),
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
107 or GTF (via gene_id/transcript id);
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
108 2. convert coordinates from GFF format--starting and ending coordinates
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
109 are 1-based, closed--to the 'traditional'/BED interval format--0 based,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
110 half-open. This is useful when using GFF files as inputs to tools that
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
111 expect traditional interval format.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
112 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
113
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
114 def __init__(self, reader, chrom_col=0, feature_col=2, start_col=3, end_col=4, strand_col=6,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
115 score_col=5, fix_strand=False, convert_to_bed_coord=False, **kwargs):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
116 NiceReaderWrapper.__init__(self, reader, chrom_col=chrom_col, start_col=start_col, end_col=end_col,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
117 strand_col=strand_col, fix_strand=fix_strand, **kwargs)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
118 self.feature_col = feature_col
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
119 self.score_col = score_col
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
120 self.convert_to_bed_coord = convert_to_bed_coord
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
121 self.last_line = None
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
122 self.cur_offset = 0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
123 self.seed_interval = None
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
124 self.seed_interval_line_len = 0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
125
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
126 def parse_row(self, line):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
127 interval = GFFInterval(self, line.split("\t"), self.chrom_col, self.feature_col, self.start_col,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
128 self.end_col, self.strand_col, self.score_col, self.default_strand,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
129 fix_strand=self.fix_strand)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
130 return interval
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
131
7
3088e7e70888 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 3
diff changeset
132 def __next__(self):
0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
133 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
134 Returns next GFFFeature.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
135 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
136
7
3088e7e70888 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 3
diff changeset
137 def handle_parse_error(e):
0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
138 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
139 Actions to take when ParseError found.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
140 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
141 if self.outstream:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
142 if self.print_delegate and hasattr(self.print_delegate, "__call__"):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
143 self.print_delegate(self.outstream, e, self)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
144 self.skipped += 1
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
145 # No reason to stuff an entire bad file into memory.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
146 if self.skipped < 10:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
147 self.skipped_lines.append((self.linenum, self.current_line, str(e)))
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
148 # Get next GFFFeature
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
149 raw_size = self.seed_interval_line_len
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
150 # If there is no seed interval, set one. Also, if there are no more
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
151 # intervals to read, this is where iterator dies.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
152 if not self.seed_interval:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
153 while not self.seed_interval:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
154 try:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
155 self.seed_interval = GenomicIntervalReader.next(self)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
156 except ParseError as e:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
157 handle_parse_error(e)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
158 finally:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
159 raw_size += len(self.current_line)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
160 # If header or comment, clear seed interval and return it with its size.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
161 if isinstance(self.seed_interval, (Header, Comment)):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
162 return_val = self.seed_interval
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
163 return_val.raw_size = len(self.current_line)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
164 self.seed_interval = None
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
165 self.seed_interval_line_len = 0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
166 return return_val
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
167 # Initialize feature identifier from seed.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
168 # For GFF.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
169 feature_group = self.seed_interval.attributes.get('group', None)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
170 # For GFF3
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
171 feature_id = self.seed_interval.attributes.get('ID', None)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
172 # For GTF.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
173 feature_transcript_id = self.seed_interval.attributes.get('transcript_id', None)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
174 # Read all intervals associated with seed.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
175 feature_intervals = []
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
176 feature_intervals.append(self.seed_interval)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
177 while True:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
178 try:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
179 interval = GenomicIntervalReader.next(self)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
180 raw_size += len(self.current_line)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
181 except StopIteration as e:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
182 # No more intervals to read, but last feature needs to be
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
183 # returned.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
184 interval = None
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
185 raw_size += len(self.current_line)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
186 break
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
187 except ParseError as e:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
188 handle_parse_error(e)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
189 raw_size += len(self.current_line)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
190 continue
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
191 # Ignore comments.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
192 if isinstance(interval, Comment):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
193 continue
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
194 # Determine if interval is part of feature.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
195 part_of = False
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
196 group = interval.attributes.get('group', None)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
197 # GFF test:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
198 if group and feature_group == group:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
199 part_of = True
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
200 # GFF3 test:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
201 parent_id = interval.attributes.get('Parent', None)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
202 cur_id = interval.attributes.get('ID', None)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
203 if (cur_id and cur_id == feature_id) or (parent_id and parent_id == feature_id):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
204 part_of = True
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
205 # GTF test:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
206 transcript_id = interval.attributes.get('transcript_id', None)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
207 if transcript_id and transcript_id == feature_transcript_id:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
208 part_of = True
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
209 # If interval is not part of feature, clean up and break.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
210 if not part_of:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
211 # Adjust raw size because current line is not part of feature.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
212 raw_size -= len(self.current_line)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
213 break
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
214 # Interval associated with feature.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
215 feature_intervals.append(interval)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
216 # Last interval read is the seed for the next interval.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
217 self.seed_interval = interval
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
218 self.seed_interval_line_len = len(self.current_line)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
219 # Return feature.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
220 feature = GFFFeature(self, self.chrom_col, self.feature_col, self.start_col,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
221 self.end_col, self.strand_col, self.score_col,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
222 self.default_strand, fix_strand=self.fix_strand,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
223 intervals=feature_intervals, raw_size=raw_size)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
224 # Convert to BED coords?
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
225 if self.convert_to_bed_coord:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
226 convert_gff_coords_to_bed(feature)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
227 return feature
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
228
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
229
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
230 def convert_bed_coords_to_gff(interval):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
231 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
232 Converts an interval object's coordinates from BED format to GFF format.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
233 Accepted object types include GenomicInterval and list (where the first
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
234 element in the list is the interval's start, and the second element is
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
235 the interval's end).
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
236 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
237 if isinstance(interval, GenomicInterval):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
238 interval.start += 1
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
239 if isinstance(interval, GFFFeature):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
240 for subinterval in interval.intervals:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
241 convert_bed_coords_to_gff(subinterval)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
242 elif isinstance(interval, list):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
243 interval[0] += 1
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
244 return interval
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
245
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
246
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
247 def convert_gff_coords_to_bed(interval):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
248 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
249 Converts an interval object's coordinates from GFF format to BED format.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
250 Accepted object types include GFFFeature, GenomicInterval, and list (where
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
251 the first element in the list is the interval's start, and the second
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
252 element is the interval's end).
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
253 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
254 if isinstance(interval, GenomicInterval):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
255 interval.start -= 1
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
256 if isinstance(interval, GFFFeature):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
257 for subinterval in interval.intervals:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
258 convert_gff_coords_to_bed(subinterval)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
259 elif isinstance(interval, list):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
260 interval[0] -= 1
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
261 return interval
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
262
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
263
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
264 def convert_to_twobit(reference_genome):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
265 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
266 Create 2bit file history fasta dataset.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
267 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
268 try:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
269 seq_path = tempfile.NamedTemporaryFile(dir=".").name
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
270 cmd = "faToTwoBit %s %s" % (reference_genome, seq_path)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
271 tmp_name = tempfile.NamedTemporaryFile(dir=".").name
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
272 tmp_stderr = open(tmp_name, 'wb')
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
273 proc = subprocess.Popen(args=cmd, shell=True, stderr=tmp_stderr.fileno())
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
274 returncode = proc.wait()
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
275 tmp_stderr.close()
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
276 if returncode != 0:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
277 # Get stderr, allowing for case where it's very large.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
278 tmp_stderr = open(tmp_name, 'rb')
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
279 stderr = ''
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
280 buffsize = 1048576
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
281 try:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
282 while True:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
283 stderr += tmp_stderr.read(buffsize)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
284 if not stderr or len(stderr) % buffsize != 0:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
285 break
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
286 except OverflowError:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
287 pass
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
288 tmp_stderr.close()
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
289 os.remove(tmp_name)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
290 stop_err(stderr)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
291 return seq_path
3
b71579ad576c planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit bd4fbe120288bf8452e479cbd82aa1bbf5c4bd31
iuc
parents: 2
diff changeset
292 except Exception as e:
0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
293 stop_err('Error running faToTwoBit. ' + str(e))
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
294
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
295
2
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
296 def get_bedtools_getfasta_default_header(chrom, start, end, strand, includes_strand_col):
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
297 """
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
298 Return a fasta header that is the default produced by the bedtools
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
299 getfasta tool, assuming "force strandedness". This will produce a
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
300 header with this format: <chrom>:<start>-<end>(strand). If the input
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
301 data includes a strand column and the strand is '+' or '-', then use it.
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
302 If the input data includes a strand column and the value of strand is
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
303 anything but '+' or '-', set strand to '.' in the header. If the input
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
304 data does not include a strand column, set strand to '.' in the header.
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
305 """
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
306 if includes_strand_col and strand in ['+', '-']:
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
307 strand_val = strand
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
308 else:
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
309 strand_val = '.'
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
310 return '%s:%s-%s(%s)' % (chrom, start, end, strand_val)
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
311
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
312
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
313 def get_fasta_header_delimiter(delimiter):
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
314 # Return a specified fasta header delimiter.
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
315 if delimiter == 'underscore':
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
316 return '_'
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
317 if delimiter == 'semicolon':
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
318 return ';'
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
319 if delimiter == 'comma':
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
320 return ','
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
321 if delimiter == 'tilde':
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
322 return '~'
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
323 if delimiter == 'vertical_bar':
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
324 return '|'
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
325 # Set the default to underscore.
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
326 return '_'
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
327
702970e4a134 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 9192c1e90e2fd5017e6044884bcc6f2e80ba8b31
iuc
parents: 0
diff changeset
328
0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
329 def get_lines(feature):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
330 # Get feature's line(s).
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
331 if isinstance(feature, GFFFeature):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
332 return feature.lines()
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
333 else:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
334 return [feature.rstrip('\r\n')]
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
335
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
336
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
337 def gff_attributes_to_str(attrs, gff_format):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
338 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
339 Convert GFF attributes to string. Supported formats are GFF3, GTF.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
340 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
341 if gff_format == 'GTF':
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
342 format_string = '%s "%s"'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
343 # Convert group (GFF) and ID, parent (GFF3) attributes to
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
344 # transcript_id, gene_id.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
345 id_attr = None
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
346 if 'group' in attrs:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
347 id_attr = 'group'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
348 elif 'ID' in attrs:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
349 id_attr = 'ID'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
350 elif 'Parent' in attrs:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
351 id_attr = 'Parent'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
352 if id_attr:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
353 attrs['transcript_id'] = attrs['gene_id'] = attrs[id_attr]
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
354 elif gff_format == 'GFF3':
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
355 format_string = '%s=%s'
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
356 attrs_strs = []
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
357 for name, value in attrs.items():
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
358 attrs_strs.append(format_string % (name, value))
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
359 return " ; ".join(attrs_strs)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
360
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
361
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
362 def parse_cols_arg(cols):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
363 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
364 Parse a columns command line argument into a four-tuple.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
365 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
366 if cols:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
367 # Handle case where no strand column included - in this case, cols
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
368 # looks something like 1,2,3,
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
369 if cols.endswith(','):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
370 cols += '0'
7
3088e7e70888 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 3
diff changeset
371 col_list = [int(x) - 1 for x in cols.split(",")]
0
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
372 return col_list
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
373 else:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
374 return BED_DEFAULT_COLS
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
375
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
376
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
377 def parse_gff_attributes(attr_str):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
378 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
379 Parses a GFF/GTF attribute string and returns a dictionary of name-value
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
380 pairs. The general format for a GFF3 attributes string is
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
381 name1=value1;name2=value2
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
382 The general format for a GTF attribute string is
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
383 name1 "value1" ; name2 "value2"
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
384 The general format for a GFF attribute string is a single string that
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
385 denotes the interval's group; in this case, method returns a dictionary
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
386 with a single key-value pair, and key name is 'group'.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
387 """
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
388 attributes_list = attr_str.split(";")
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
389 attributes = {}
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
390 for name_value_pair in attributes_list:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
391 # Try splitting by '=' (GFF3) first because spaces are allowed in GFF3
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
392 # attribute; next, try double quotes for GTF.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
393 pair = name_value_pair.strip().split("=")
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
394 if len(pair) == 1:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
395 pair = name_value_pair.strip().split("\"")
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
396 if len(pair) == 1:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
397 # Could not split for some reason.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
398 continue
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
399 if pair == '':
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
400 continue
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
401 name = pair[0].strip()
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
402 if name == '':
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
403 continue
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
404 # Need to strip double quote from values
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
405 value = pair[1].strip(" \"")
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
406 attributes[name] = value
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
407 if len(attributes) == 0:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
408 # Could not split attributes string, so entire string must be
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
409 # 'group' attribute. This is the case for strictly GFF files.
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
410 attributes['group'] = attr_str
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
411 return attributes
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
412
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
413
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
414 def reverse_complement(s):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
415 complement_dna = {"A": "T", "T": "A", "C": "G", "G": "C", "a": "t", "t": "a", "c": "g", "g": "c", "N": "N", "n": "n"}
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
416 reversed_s = []
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
417 for i in s:
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
418 reversed_s.append(complement_dna[i])
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
419 reversed_s.reverse()
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
420 return "".join(reversed_s)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
421
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
422
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
423 def stop_err(msg):
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
424 sys.stderr.write(msg)
8dd8e89c0603 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/extract_genomic_dna commit b'67cff25a50ba173b0468819204d0999496f68ea9'
iuc
parents:
diff changeset
425 sys.exit(1)