Mercurial > repos > fubar > jbrowse2

annotate jb2_GFF/GFFParser.py @ 18:2e6c48910819 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a572525c0f1d7b4dbae1c3aaa4e748aa019d8347
author fubar
date Mon, 29 Jan 2024 02:34:43 +0000
parents 4c201a3d4755
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
17
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1 """Parse GFF files into features attached to Biopython SeqRecord objects.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
2
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
3 This deals with GFF3 formatted files, a tab delimited format for storing
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
4 sequence features and annotations:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
5
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
6 http://www.sequenceontology.org/gff3.shtml
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
7
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
8 It will also deal with older GFF versions (GTF/GFF2):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
9
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
10 http://www.sanger.ac.uk/Software/formats/GFF/GFF_Spec.shtml
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
11 http://mblab.wustl.edu/GTF22.html
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
12
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
13 The implementation utilizes map/reduce parsing of GFF using Disco. Disco
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
14 (http://discoproject.org) is a Map-Reduce framework for Python utilizing
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
15 Erlang for parallelization. The code works on a single processor without
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
16 Disco using the same architecture.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
17 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
18 import os
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
19 import copy
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
20 import json
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
21 import re
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
22 import collections
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
23 import io
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
24 import itertools
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
25 import warnings
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
26 import six
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
27 from six.moves import urllib
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
28
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
29 from Bio.SeqRecord import SeqRecord
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
30 from Bio import SeqFeature
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
31 from Bio import SeqIO
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
32 from Bio import BiopythonDeprecationWarning
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
33
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
34 import disco
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
35
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
36 # Make defaultdict compatible with versions of python older than 2.4
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
37 try:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
38 collections.defaultdict
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
39 except AttributeError:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
40 import _utils
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
41
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
42 collections.defaultdict = _utils.defaultdict
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
43
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
44 unknown_seq_avail = False
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
45 try:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
46 from Bio.Seq import UnknownSeq
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
47
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
48 unknown_seq_avail = True
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
49 except ImportError:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
50 # Starting with biopython 1.81, has been removed
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
51 from Bio.Seq import _UndefinedSequenceData
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
52 from Bio.Seq import Seq
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
53
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
54
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
55 warnings.simplefilter("ignore", BiopythonDeprecationWarning)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
56
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
57
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
58 def _gff_line_map(line, params):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
59 """Map part of Map-Reduce; parses a line of GFF into a dictionary.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
60
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
61 Given an input line from a GFF file, this:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
62 - decides if the file passes our filtering limits
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
63 - if so:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
64 - breaks it into component elements
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
65 - determines the type of attribute (flat, parent, child or annotation)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
66 - generates a dictionary of GFF info which can be serialized as JSON
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
67 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
68
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
69 def _merge_keyvals(parts):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
70 """Merge key-values escaped by quotes
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
71 that are improperly split at semicolons."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
72 out = []
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
73 for i, p in enumerate(parts):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
74 if (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
75 i > 0
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
76 and len(p) == 1
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
77 and p[0].endswith('"')
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
78 and not p[0].startswith('"')
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
79 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
80 if out[-1][-1].startswith('"'):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
81 prev_p = out.pop(-1)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
82 to_merge = prev_p[-1]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
83 prev_p[-1] = "%s; %s" % (to_merge, p[0])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
84 out.append(prev_p)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
85 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
86 out.append(p)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
87 return out
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
88
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
89 gff3_kw_pat = re.compile(r"\w+=")
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
90
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
91 def _split_keyvals(keyval_str):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
92 """Split key-value pairs in a GFF2, GTF and GFF3 compatible way.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
93
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
94 GFF3 has key value pairs like:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
95 count=9;gene=amx-2;sequence=SAGE:aacggagccg
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
96 GFF2 and GTF have:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
97 Sequence "Y74C9A" ; Note "Clone Y74C9A; Genbank AC024206"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
98 name "fgenesh1_pg.C_chr_1000003"; transcriptId 869
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
99 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
100 quals = collections.defaultdict(list)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
101 if keyval_str is None:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
102 return quals
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
103 # ensembl GTF has a stray semi-colon at the end
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
104 if keyval_str[-1] == ";":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
105 keyval_str = keyval_str[:-1]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
106 # GFF2/GTF has a semi-colon with at least one space after it.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
107 # It can have spaces on both sides; wormbase does this.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
108 # GFF3 works with no spaces.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
109 # Split at the first one we can recognize as working
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
110 parts = keyval_str.split(" ; ")
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
111 if len(parts) == 1:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
112 parts = [x.strip() for x in keyval_str.split(";")]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
113 # check if we have GFF3 style key-vals (with =)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
114 is_gff2 = True
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
115 if gff3_kw_pat.match(parts[0]):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
116 is_gff2 = False
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
117 key_vals = _merge_keyvals([p.split("=") for p in parts])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
118 # otherwise, we are separated by a space with a key as the first item
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
119 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
120 pieces = []
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
121 for p in parts:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
122 # fix misplaced semi-colons in keys in some GFF2 files
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
123 if p and p[0] == ";":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
124 p = p[1:]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
125 pieces.append(p.strip().split(" "))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
126 key_vals = [(p[0], " ".join(p[1:])) for p in pieces]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
127 for item in key_vals:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
128 # standard in-spec items are key=value
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
129 if len(item) == 2:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
130 key, val = item
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
131 # out-of-spec files can have just key values. We set an empty value
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
132 # which will be changed to true later to standardize.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
133 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
134 assert len(item) == 1, item
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
135 key = item[0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
136 val = ""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
137 # remove quotes in GFF2 files
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
138 quoted = False
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
139 if len(val) > 0 and val[0] == '"' and val[-1] == '"':
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
140 quoted = True
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
141 val = val[1:-1]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
142 if val:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
143 if quoted:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
144 quals[key].append(val)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
145 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
146 quals[key].extend(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
147 [v for v in val.split(",") if v]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
148 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
149 # if we don't have a value, make this a key=True/False style
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
150 # attribute
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
151 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
152 quals[key].append("true")
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
153 for key, vals in quals.items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
154 quals[key] = [urllib.parse.unquote(v) for v in vals]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
155 return quals, is_gff2
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
156
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
157 def _nest_gff2_features(gff_parts):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
158 """Provide nesting of GFF2 transcript parts with transcript IDs.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
159
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
160 exons and coding sequences are mapped to a parent with a transcript_id
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
161 in GFF2. This is implemented differently at different genome centers
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
162 and this function attempts to resolve that and map things to the GFF3
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
163 way of doing them.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
164 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
165 # map protein or transcript ids to a parent
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
166 for transcript_id in [
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
167 "transcript_id",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
168 "transcriptId",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
169 "proteinId",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
170 ]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
171 try:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
172 gff_parts["quals"]["Parent"] = gff_parts["quals"][
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
173 transcript_id
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
174 ]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
175 break
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
176 except KeyError:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
177 pass
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
178 # case for WormBase GFF -- everything labelled as Transcript or CDS
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
179 for flat_name in ["Transcript", "CDS"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
180 if flat_name in gff_parts["quals"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
181 # parent types
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
182 if gff_parts["type"] in [flat_name]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
183 if not gff_parts["id"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
184 gff_parts["id"] = gff_parts["quals"][
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
185 flat_name
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
186 ][0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
187 gff_parts["quals"]["ID"] = [gff_parts["id"]]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
188 # children types
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
189 elif gff_parts["type"] in [
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
190 "intron",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
191 "exon",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
192 "three_prime_UTR",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
193 "coding_exon",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
194 "five_prime_UTR",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
195 "CDS",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
196 "stop_codon",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
197 "start_codon",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
198 ]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
199 gff_parts["quals"]["Parent"] = gff_parts["quals"][
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
200 flat_name
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
201 ]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
202 break
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
203
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
204 return gff_parts
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
205
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
206 strand_map = {"+": 1, "-": -1, "?": None, None: None}
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
207 line = line.strip()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
208 if line[:2] == "##":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
209 return [("directive", line[2:])]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
210 elif line and line[0] != "#":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
211 parts = line.split("\t")
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
212 should_do = True
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
213 if params.limit_info:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
214 for limit_name, limit_values in params.limit_info.items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
215 cur_id = tuple(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
216 [parts[i] for i in params.filter_info[limit_name]]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
217 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
218 if cur_id not in limit_values:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
219 should_do = False
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
220 break
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
221 if should_do:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
222 assert len(parts) >= 8, line
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
223 # not python2.4 compatible but easier to understand
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
224 # gff_parts = [(None if p == '.' else p) for p in parts]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
225 gff_parts = []
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
226 for p in parts:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
227 if p == ".":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
228 gff_parts.append(None)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
229 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
230 gff_parts.append(p)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
231 gff_info = dict()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
232 # collect all of the base qualifiers for this item
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
233 if len(parts) > 8:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
234 quals, is_gff2 = _split_keyvals(gff_parts[8])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
235 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
236 quals, is_gff2 = collections.defaultdict(list), False
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
237 gff_info["is_gff2"] = is_gff2
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
238 if gff_parts[1]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
239 quals["source"].append(gff_parts[1])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
240 if gff_parts[5]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
241 quals["score"].append(gff_parts[5])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
242 if gff_parts[7]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
243 quals["phase"].append(gff_parts[7])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
244 gff_info["quals"] = dict(quals)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
245 gff_info["rec_id"] = gff_parts[0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
246 # if we are describing a location, then we are a feature
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
247 if gff_parts[3] and gff_parts[4]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
248 gff_info["location"] = [
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
249 int(gff_parts[3]) - 1,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
250 int(gff_parts[4]),
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
251 ]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
252 gff_info["type"] = gff_parts[2]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
253 gff_info["id"] = quals.get("ID", [""])[0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
254 gff_info["strand"] = strand_map.get(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
255 gff_parts[6], None
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
256 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
257 if is_gff2:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
258 gff_info = _nest_gff2_features(gff_info)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
259 # features that have parents need to link so we can pick up
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
260 # the relationship
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
261 if "Parent" in gff_info["quals"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
262 # check for self referential parent/child relationships
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
263 # remove the ID, which is not useful
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
264 for p in gff_info["quals"]["Parent"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
265 if p == gff_info["id"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
266 gff_info["id"] = ""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
267 del gff_info["quals"]["ID"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
268 break
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
269 final_key = "child"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
270 elif gff_info["id"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
271 final_key = "parent"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
272 # Handle flat features
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
273 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
274 final_key = "feature"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
275 # otherwise, associate these annotations with the full record
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
276 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
277 final_key = "annotation"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
278 if params.jsonify:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
279 return [(final_key, json.dumps(gff_info))]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
280 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
281 return [(final_key, gff_info)]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
282 return []
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
283
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
284
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
285 def _gff_line_reduce(map_results, out, params):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
286 """Reduce part of Map-Reduce; combines results of parsed features."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
287 final_items = dict()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
288 for gff_type, final_val in map_results:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
289 if params.jsonify and gff_type not in ["directive"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
290 final_val = json.loads(final_val)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
291 try:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
292 final_items[gff_type].append(final_val)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
293 except KeyError:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
294 final_items[gff_type] = [final_val]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
295 for key, vals in final_items.items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
296 if params.jsonify:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
297 vals = json.dumps(vals)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
298 out.add(key, vals)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
299
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
300
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
301 class _MultiIDRemapper:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
302 """Provide an ID remapping for cases where a parent has a non-unique ID.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
303
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
304 Real life GFF3 cases have non-unique ID attributes, which we fix here
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
305 by using the unique sequence region to assign children to the right
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
306 parent.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
307 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
308
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
309 def __init__(self, base_id, all_parents):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
310 self._base_id = base_id
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
311 self._parents = all_parents
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
312
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
313 def remap_id(self, feature_dict):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
314 rstart, rend = feature_dict["location"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
315 for index, parent in enumerate(self._parents):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
316 pstart, pend = parent["location"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
317 if rstart >= pstart and rend <= pend:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
318 if index > 0:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
319 return "%s_%s" % (self._base_id, index + 1)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
320 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
321 return self._base_id
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
322 # if we haven't found a location match but parents are umabiguous,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
323 # return that
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
324 if len(self._parents) == 1:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
325 return self._base_id
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
326 raise ValueError(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
327 "Did not find remapped ID location: %s, %s, %s"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
328 % (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
329 self._base_id,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
330 [p["location"] for p in self._parents],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
331 feature_dict["location"],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
332 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
333 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
334
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
335
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
336 class _AbstractMapReduceGFF:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
337 """Base class providing general GFF parsing for local and remote classes.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
338
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
339 This class should be subclassed to provide a concrete class to parse
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
340 GFF under specific conditions. These classes need to implement
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
341 the _gff_process function, which returns a dictionary of SeqRecord
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
342 information.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
343 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
344
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
345 def __init__(self, create_missing=True):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
346 """Initialize GFF parser
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
347
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
348 create_missing - If True, create blank records for GFF ids not in
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
349 the base_dict. If False, an error will be raised.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
350 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
351 self._create_missing = create_missing
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
352 self._map_fn = _gff_line_map
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
353 self._reduce_fn = _gff_line_reduce
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
354 self._examiner = GFFExaminer()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
355
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
356 def _gff_process(self, gff_files, limit_info, target_lines=None):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
357 raise NotImplementedError("Derived class must define")
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
358
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
359 def parse(self, gff_files, base_dict=None, limit_info=None):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
360 """Parse a GFF file, returning an iterator of SeqRecords.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
361
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
362 limit_info - A dictionary specifying the regions of the GFF file
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
363 which should be extracted. This allows only relevant portions of a file
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
364 to be parsed.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
365
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
366 base_dict - A base dictionary of SeqRecord objects which may be
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
367 pre-populated with sequences and other features. The new features from
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
368 the GFF file will be added to this dictionary.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
369 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
370 for rec in self.parse_in_parts(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
371 gff_files, base_dict, limit_info
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
372 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
373 yield rec
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
374
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
375 def parse_in_parts(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
376 self,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
377 gff_files,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
378 base_dict=None,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
379 limit_info=None,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
380 target_lines=None,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
381 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
382 """Parse a region of a GFF file specified, returning info as generated.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
383
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
384 target_lines -- The number of lines in the file which should be used
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
385 for each partial parse. This should be determined based on available
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
386 memory.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
387 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
388 for results in self.parse_simple(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
389 gff_files, limit_info, target_lines
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
390 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
391 if base_dict is None:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
392 cur_dict = dict()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
393 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
394 cur_dict = copy.deepcopy(base_dict)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
395 cur_dict = self._results_to_features(cur_dict, results)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
396 all_ids = list(cur_dict.keys())
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
397 all_ids.sort()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
398 for cur_id in all_ids:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
399 yield cur_dict[cur_id]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
400
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
401 def parse_simple(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
402 self, gff_files, limit_info=None, target_lines=1
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
403 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
404 """Simple parse which does not build or nest features.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
405
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
406 This returns a simple dictionary representation of each line in the
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
407 GFF file.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
408 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
409 # gracefully handle a single file passed
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
410 if not isinstance(gff_files, (list, tuple)):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
411 gff_files = [gff_files]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
412 limit_info = self._normalize_limit_info(limit_info)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
413 for results in self._gff_process(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
414 gff_files, limit_info, target_lines
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
415 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
416 yield results
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
417
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
418 def _normalize_limit_info(self, limit_info):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
419 """Turn all limit information into tuples for identical comparisons."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
420 final_limit_info = {}
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
421 if limit_info:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
422 for key, values in limit_info.items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
423 final_limit_info[key] = []
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
424 for v in values:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
425 if isinstance(v, str):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
426 final_limit_info[key].append((v,))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
427 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
428 final_limit_info[key].append(tuple(v))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
429 return final_limit_info
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
430
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
431 def _results_to_features(self, base, results):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
432 """Add parsed dictionaries of results to Biopython SeqFeatures."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
433 base = self._add_annotations(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
434 base, results.get("annotation", [])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
435 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
436 for feature in results.get("feature", []):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
437 (_, base) = self._add_toplevel_feature(base, feature)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
438 base = self._add_parent_child_features(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
439 base, results.get("parent", []), results.get("child", [])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
440 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
441 base = self._add_seqs(base, results.get("fasta", []))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
442 base = self._add_directives(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
443 base, results.get("directive", [])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
444 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
445 return base
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
446
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
447 def _add_directives(self, base, directives):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
448 """Handle any directives or meta-data in the GFF file.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
449
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
450 Relevant items are added as annotation meta-data to each record.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
451 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
452 dir_keyvals = collections.defaultdict(list)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
453 for directive in directives:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
454 parts = directive.split()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
455 if len(parts) > 1:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
456 key = parts[0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
457 if len(parts) == 2:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
458 val = parts[1]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
459 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
460 val = tuple(parts[1:])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
461 # specific directives that need special handling
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
462 if (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
463 key == "sequence-region"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
464 ): # convert to Python 0-based coordinates
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
465 if len(val) == 2: # handle regions missing contig
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
466 val = (int(val[0]) - 1, int(val[1]))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
467 elif len(val) == 3:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
468 val = (val[0], int(val[1]) - 1, int(val[2]))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
469 dir_keyvals[key].append(val)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
470 for key, vals in dir_keyvals.items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
471 for rec in base.values():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
472 self._add_ann_to_rec(rec, key, vals)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
473 return base
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
474
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
475 def _get_matching_record_id(self, base, find_id):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
476 """Find a matching base record with the test identifier, handling tricky cases.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
477
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
478 NCBI IDs https://en.wikipedia.org/wiki/FASTA_format#NCBI_identifiers
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
479 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
480 # Straight matches for identifiers
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
481 if find_id in base:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
482 return find_id
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
483 # NCBI style IDs in find_id
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
484 elif find_id and find_id.find("|") > 0:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
485 for test_id in [
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
486 x.strip() for x in find_id.split("|")[1:]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
487 ]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
488 if test_id and test_id in base:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
489 return test_id
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
490 # NCBI style IDs in base IDs
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
491 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
492 for base_id in base.keys():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
493 if base_id.find("|") > 0:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
494 for test_id in [
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
495 x.strip() for x in base_id.split("|")[1:]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
496 ]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
497 if test_id and test_id == find_id:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
498 return base_id
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
499 return None
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
500
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
501 def _add_seqs(self, base, recs):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
502 """Add sequence information contained in the GFF3 to records."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
503 for rec in recs:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
504 match_id = self._get_matching_record_id(base, rec.id)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
505 if match_id:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
506 base[match_id].seq = rec.seq
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
507 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
508 base[rec.id] = rec
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
509 return base
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
510
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
511 def _add_parent_child_features(self, base, parents, children):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
512 """Add nested features with parent child relationships."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
513 multi_remap = self._identify_dup_ids(parents)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
514 # add children features
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
515 children_prep = collections.defaultdict(list)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
516 for child_dict in children:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
517 child_feature = self._get_feature(child_dict)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
518 for pindex, pid in enumerate(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
519 child_feature.qualifiers["Parent"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
520 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
521 if pid in multi_remap:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
522 pid = multi_remap[pid].remap_id(child_dict)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
523 child_feature.qualifiers["Parent"][pindex] = pid
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
524 children_prep[pid].append(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
525 (child_dict["rec_id"], child_feature)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
526 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
527 children = dict(children_prep)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
528 # add children to parents that exist
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
529 for cur_parent_dict in parents:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
530 cur_id = cur_parent_dict["id"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
531 if cur_id in multi_remap:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
532 cur_parent_dict["id"] = multi_remap[cur_id].remap_id(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
533 cur_parent_dict
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
534 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
535 cur_parent, base = self._add_toplevel_feature(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
536 base, cur_parent_dict
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
537 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
538 cur_parent, children = self._add_children_to_parent(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
539 cur_parent, children
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
540 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
541 # create parents for children without them (GFF2 or split/bad files)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
542 while len(children) > 0:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
543 parent_id, cur_children = next(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
544 itertools.islice(children.items(), 1)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
545 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
546 # one child, do not nest it
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
547 if len(cur_children) == 1:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
548 rec_id, child = cur_children[0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
549 loc = (child.location.start, child.location.end)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
550 rec, base = self._get_rec(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
551 base, dict(rec_id=rec_id, location=loc)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
552 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
553 rec.features.append(child)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
554 del children[parent_id]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
555 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
556 cur_parent, base = self._add_missing_parent(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
557 base, parent_id, cur_children
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
558 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
559 cur_parent, children = self._add_children_to_parent(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
560 cur_parent, children
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
561 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
562 return base
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
563
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
564 def _identify_dup_ids(self, parents):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
565 """Identify duplicated ID attributes in potential nested parents.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
566
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
567 According to the GFF3 spec ID attributes are supposed to be unique
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
568 for a file, but this is not always true in practice. This looks
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
569 for duplicates, and provides unique IDs sorted by locations.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
570 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
571 multi_ids = collections.defaultdict(list)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
572 for parent in parents:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
573 multi_ids[parent["id"]].append(parent)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
574 multi_ids = [
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
575 (mid, ps)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
576 for (mid, ps) in multi_ids.items()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
577 if len(parents) > 1
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
578 ]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
579 multi_remap = dict()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
580 for mid, parents in multi_ids:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
581 multi_remap[mid] = _MultiIDRemapper(mid, parents)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
582 return multi_remap
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
583
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
584 def _add_children_to_parent(self, cur_parent, children):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
585 """Recursively add children to parent features."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
586 if cur_parent.id in children:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
587 cur_children = children[cur_parent.id]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
588 ready_children = []
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
589 for _, cur_child in cur_children:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
590 cur_child, _ = self._add_children_to_parent(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
591 cur_child, children
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
592 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
593 ready_children.append(cur_child)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
594 # Support Biopython features for 1.62+
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
595 # CompoundLocations and pre-1.62
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
596 if not hasattr(SeqFeature, "CompoundLocation"):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
597 cur_parent.location_operator = "join"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
598 for cur_child in ready_children:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
599 cur_parent.sub_features.append(cur_child)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
600 del children[cur_parent.id]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
601 return cur_parent, children
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
602
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
603 def _add_annotations(self, base, anns):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
604 """Add annotation data from the GFF file to records."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
605 # add these as a list of annotations, checking not to overwrite
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
606 # current values
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
607 for ann in anns:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
608 rec, base = self._get_rec(base, ann)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
609 for key, vals in ann["quals"].items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
610 self._add_ann_to_rec(rec, key, vals)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
611 return base
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
612
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
613 def _add_ann_to_rec(self, rec, key, vals):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
614 """Add a key/value annotation to the given SeqRecord."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
615 if key in rec.annotations:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
616 try:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
617 rec.annotations[key].extend(vals)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
618 except AttributeError:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
619 rec.annotations[key] = [rec.annotations[key]] + vals
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
620 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
621 rec.annotations[key] = vals
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
622
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
623 def _get_rec(self, base, info_dict):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
624 """Retrieve a record to add features to."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
625 max_loc = info_dict.get("location", (0, 1))[1]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
626 match_id = self._get_matching_record_id(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
627 base, info_dict["rec_id"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
628 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
629 if match_id:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
630 cur_rec = base[match_id]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
631 # update generated unknown sequences
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
632 if unknown_seq_avail and isinstance(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
633 cur_rec.seq, UnknownSeq
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
634 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
635 cur_rec.seq._length = max(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
636 [max_loc, cur_rec.seq._length]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
637 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
638 elif not unknown_seq_avail and isinstance(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
639 cur_rec.seq._data, _UndefinedSequenceData
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
640 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
641 cur_rec.seq._data._length = max(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
642 [max_loc, cur_rec.seq._data._length]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
643 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
644 return cur_rec, base
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
645 elif self._create_missing:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
646 if unknown_seq_avail:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
647 new_rec = SeqRecord(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
648 UnknownSeq(max_loc), info_dict["rec_id"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
649 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
650 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
651 new_rec = SeqRecord(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
652 Seq(None, length=max_loc), info_dict["rec_id"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
653 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
654 base[info_dict["rec_id"]] = new_rec
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
655 return new_rec, base
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
656 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
657 raise KeyError(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
658 "Did not find matching record in %s for %s"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
659 % (base.keys(), info_dict)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
660 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
661
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
662 def _add_missing_parent(self, base, parent_id, cur_children):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
663 """Add a new feature that is missing from the GFF file."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
664 base_rec_id = list(set(c[0] for c in cur_children))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
665 child_strands = list(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
666 set(c[1].location.strand for c in cur_children)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
667 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
668 inferred_strand = (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
669 child_strands[0] if len(child_strands) == 1 else None
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
670 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
671 assert len(base_rec_id) > 0
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
672 feature_dict = dict(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
673 id=parent_id,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
674 strand=inferred_strand,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
675 type="inferred_parent",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
676 quals=dict(ID=[parent_id]),
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
677 rec_id=base_rec_id[0],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
678 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
679 coords = [
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
680 (c.location.start, c.location.end)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
681 for r, c in cur_children
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
682 ]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
683 feature_dict["location"] = (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
684 min([c[0] for c in coords]),
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
685 max([c[1] for c in coords]),
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
686 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
687 return self._add_toplevel_feature(base, feature_dict)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
688
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
689 def _add_toplevel_feature(self, base, feature_dict):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
690 """Add a toplevel non-nested feature to the appropriate record."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
691 new_feature = self._get_feature(feature_dict)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
692 rec, base = self._get_rec(base, feature_dict)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
693 rec.features.append(new_feature)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
694 return new_feature, base
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
695
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
696 def _get_feature(self, feature_dict):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
697 """Retrieve a Biopython feature from our dictionary representation."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
698 # location = SeqFeature.FeatureLocation(*feature_dict['location'])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
699 rstart, rend = feature_dict["location"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
700 new_feature = SeqFeature.SeqFeature(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
701 SeqFeature.SimpleLocation(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
702 start=rstart, end=rend, strand=feature_dict["strand"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
703 ),
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
704 feature_dict["type"],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
705 id=feature_dict["id"],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
706 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
707 # Support for Biopython 1.68 and above, which removed sub_features
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
708 if not hasattr(new_feature, "sub_features"):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
709 new_feature.sub_features = []
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
710 new_feature.qualifiers = feature_dict["quals"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
711 return new_feature
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
712
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
713 def _parse_fasta(self, in_handle):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
714 """Parse FASTA sequence information contained in the GFF3 file."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
715 return list(SeqIO.parse(in_handle, "fasta"))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
716
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
717
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
718 class _GFFParserLocalOut:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
719 """Provide a collector for local GFF MapReduce file parsing."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
720
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
721 def __init__(self, smart_breaks=False):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
722 self._items = dict()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
723 self._smart_breaks = smart_breaks
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
724 self._missing_keys = collections.defaultdict(int)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
725 self._last_parent = None
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
726 self.can_break = True
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
727 self.num_lines = 0
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
728
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
729 def add(self, key, vals):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
730 if self._smart_breaks:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
731 # if we are not GFF2 we expect parents and break
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
732 # based on not having missing ones
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
733 if key == "directive":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
734 if vals[0] == "#":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
735 self.can_break = True
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
736 self._last_parent = None
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
737 elif not vals[0].get("is_gff2", False):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
738 self._update_missing_parents(key, vals)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
739 self.can_break = len(self._missing_keys) == 0
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
740 # break when we are done with stretches of child features
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
741 elif key != "child":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
742 self.can_break = True
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
743 self._last_parent = None
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
744 # break when we have lots of child features in a row
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
745 # and change between parents
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
746 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
747 cur_parent = vals[0]["quals"]["Parent"][0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
748 if self._last_parent:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
749 self.can_break = cur_parent != self._last_parent
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
750 self._last_parent = cur_parent
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
751 self.num_lines += 1
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
752 try:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
753 self._items[key].extend(vals)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
754 except KeyError:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
755 self._items[key] = vals
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
756
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
757 def _update_missing_parents(self, key, vals):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
758 # smart way of deciding if we can break this.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
759 # if this is too much, can go back to not breaking in the
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
760 # middle of children
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
761 if key in ["child"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
762 for val in vals:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
763 for p_id in val["quals"]["Parent"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
764 self._missing_keys[p_id] += 1
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
765 for val in vals:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
766 try:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
767 del self._missing_keys[val["quals"]["ID"][0]]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
768 except KeyError:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
769 pass
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
770
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
771 def has_items(self):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
772 return len(self._items) > 0
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
773
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
774 def get_results(self):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
775 self._last_parent = None
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
776 return self._items
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
777
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
778
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
779 class GFFParser(_AbstractMapReduceGFF):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
780 """Local GFF parser providing standardized parsing of GFF files."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
781
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
782 def __init__(self, line_adjust_fn=None, create_missing=True):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
783 _AbstractMapReduceGFF.__init__(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
784 self, create_missing=create_missing
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
785 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
786 self._line_adjust_fn = line_adjust_fn
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
787
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
788 def _gff_process(self, gff_files, limit_info, target_lines):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
789 """Process GFF addition without any parallelization.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
790
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
791 In addition to limit filtering, this accepts a target_lines attribute
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
792 which provides a number of lines to parse before returning results.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
793 This allows partial parsing of a file to prevent memory issues.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
794 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
795 line_gen = self._file_line_generator(gff_files)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
796 for out in self._lines_to_out_info(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
797 line_gen, limit_info, target_lines
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
798 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
799 yield out
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
800
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
801 def _file_line_generator(self, gff_files):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
802 """Generate single lines from a set of GFF files."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
803 for gff_file in gff_files:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
804 if hasattr(gff_file, "read"):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
805 need_close = False
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
806 in_handle = gff_file
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
807 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
808 need_close = True
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
809 in_handle = open(gff_file)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
810 while 1:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
811 line = in_handle.readline()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
812 if not line:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
813 break
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
814 yield line
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
815 if need_close:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
816 in_handle.close()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
817
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
818 def _lines_to_out_info(self, line_iter, limit_info=None, target_lines=None):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
819 """Generate SeqRecord and SeqFeatures from GFF file lines."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
820 params = self._examiner._get_local_params(limit_info)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
821 out_info = _GFFParserLocalOut(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
822 (target_lines is not None and target_lines > 1)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
823 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
824 found_seqs = False
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
825 for line in line_iter:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
826 results = self._map_fn(line, params)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
827 if self._line_adjust_fn and results:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
828 if results[0][0] not in ["directive"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
829 results = [
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
830 (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
831 results[0][0],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
832 self._line_adjust_fn(results[0][1]),
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
833 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
834 ]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
835 self._reduce_fn(results, out_info, params)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
836 if (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
837 target_lines
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
838 and out_info.num_lines >= target_lines
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
839 and out_info.can_break
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
840 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
841 yield out_info.get_results()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
842 out_info = _GFFParserLocalOut(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
843 (target_lines is not None and target_lines > 1)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
844 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
845 if (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
846 results
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
847 and results[0][0] == "directive"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
848 and results[0][1] == "FASTA"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
849 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
850 found_seqs = True
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
851 break
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
852
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
853 class FakeHandle:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
854 def __init__(self, line_iter):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
855 self._iter = line_iter
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
856
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
857 def __iter__(self):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
858 return self
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
859
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
860 def __next__(self):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
861 return next(self._iter)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
862
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
863 next = __next__
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
864
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
865 def read(self, size=-1):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
866 if size < 0:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
867 return "".join(x for x in self._iter)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
868 elif size == 0:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
869 return "" # Used by Biopython to sniff unicode vs bytes
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
870 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
871 raise NotImplementedError
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
872
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
873 def readline(self):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
874 try:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
875 return next(self._iter)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
876 except StopIteration:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
877 return ""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
878
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
879 if found_seqs:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
880 fasta_recs = self._parse_fasta(FakeHandle(line_iter))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
881 out_info.add("fasta", fasta_recs)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
882 if out_info.has_items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
883 yield out_info.get_results()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
884
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
885
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
886 class DiscoGFFParser(_AbstractMapReduceGFF):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
887 """GFF Parser with parallelization (http://discoproject.org."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
888
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
889 def __init__(self, disco_host, create_missing=True):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
890 """Initialize parser.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
891
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
892 disco_host - Web reference to a Disco host which will be used for
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
893 parallelizing the GFF reading job.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
894 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
895 _AbstractMapReduceGFF.__init__(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
896 self, create_missing=create_missing
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
897 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
898 self._disco_host = disco_host
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
899
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
900 def _gff_process(self, gff_files, limit_info, target_lines=None):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
901 """Process GFF addition, using Disco to parallelize the process."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
902 assert target_lines is None, "Cannot split parallelized jobs"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
903 # make these imports local; only need them when using disco
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
904 # absolute path names unless they are special disco files
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
905 full_files = []
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
906 for f in gff_files:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
907 if f.split(":")[0] != "disco":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
908 full_files.append(os.path.abspath(f))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
909 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
910 full_files.append(f)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
911 results = disco.job(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
912 self._disco_host,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
913 name="gff_reader",
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
914 input=full_files,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
915 params=disco.Params(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
916 limit_info=limit_info,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
917 jsonify=True,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
918 filter_info=self._examiner._filter_info,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
919 ),
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
920 required_modules=["json", "collections", "re"],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
921 map=self._map_fn,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
922 reduce=self._reduce_fn,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
923 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
924 processed = dict()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
925 for out_key, out_val in disco.result_iterator(results):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
926 processed[out_key] = json.loads(out_val)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
927 yield processed
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
928
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
929
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
930 def parse(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
931 gff_files, base_dict=None, limit_info=None, target_lines=None
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
932 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
933 """parse GFF files into SeqRecords and SeqFeatures."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
934 parser = GFFParser()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
935 for rec in parser.parse_in_parts(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
936 gff_files, base_dict, limit_info, target_lines
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
937 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
938 yield rec
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
939
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
940
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
941 def parse_simple(gff_files, limit_info=None):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
942 """Parse GFF files as line by line dictionary of parts."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
943 parser = GFFParser()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
944 for rec in parser.parse_simple(gff_files, limit_info=limit_info):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
945 if "child" in rec:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
946 assert "parent" not in rec
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
947 yield rec["child"][0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
948 elif "parent" in rec:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
949 yield rec["parent"][0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
950 elif "feature" in rec:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
951 yield rec["feature"][0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
952 # ignore directive lines
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
953 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
954 assert "directive" in rec
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
955
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
956
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
957 def _file_or_handle(fn):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
958 """Decorator to handle either an input handle or a file."""
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
959
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
960 def _file_or_handle_inside(*args, **kwargs):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
961 in_file = args[1]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
962 if hasattr(in_file, "read"):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
963 need_close = False
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
964 in_handle = in_file
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
965 if six.PY3 and not isinstance(in_handle, io.TextIOBase):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
966 raise TypeError(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
967 "input handle must be opened in text mode"
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
968 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
969 else:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
970 need_close = True
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
971 in_handle = open(in_file)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
972 args = (args[0], in_handle) + args[2:]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
973 out = fn(*args, **kwargs)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
974 if need_close:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
975 in_handle.close()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
976 return out
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
977
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
978 return _file_or_handle_inside
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
979
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
980
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
981 class GFFExaminer:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
982 """Provide high level details about a GFF file to refine parsing.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
983
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
984 GFF is a spec and is provided by many different centers. Real life files
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
985 will present the same information in slightly different ways. Becoming
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
986 familiar with the file you are dealing with is the best way to extract the
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
987 information you need. This class provides high level summary details to
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
988 help in learning.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
989 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
990
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
991 def __init__(self):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
992 self._filter_info = dict(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
993 gff_id=[0],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
994 gff_source_type=[1, 2],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
995 gff_source=[1],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
996 gff_type=[2],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
997 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
998
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
999 def _get_local_params(self, limit_info=None):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1000 class _LocalParams:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1001 def __init__(self):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1002 self.jsonify = False
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1003
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1004 params = _LocalParams()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1005 params.limit_info = limit_info
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1006 params.filter_info = self._filter_info
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1007 return params
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1008
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1009 @_file_or_handle
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1010 def available_limits(self, gff_handle):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1011 """Return dictionary information on possible limits for this file.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1012
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1013 This returns a nested dictionary with the following structure:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1014
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1015 keys -- names of items to filter by
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1016 values -- dictionary with:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1017 keys -- filter choice
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1018 value -- counts of that filter in this file
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1019
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1020 Not a parallelized map-reduce implementation.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1021 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1022 cur_limits = dict()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1023 for filter_key in self._filter_info.keys():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1024 cur_limits[filter_key] = collections.defaultdict(int)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1025 for line in gff_handle:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1026 # when we hit FASTA sequences, we are done with annotations
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1027 if line.startswith("##FASTA"):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1028 break
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1029 # ignore empty and comment lines
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1030 if line.strip() and line.strip()[0] != "#":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1031 parts = [p.strip() for p in line.split("\t")]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1032 assert len(parts) >= 8, line
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1033 parts = parts[:9]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1034 for (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1035 filter_key,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1036 cur_indexes,
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1037 ) in self._filter_info.items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1038 cur_id = tuple([parts[i] for i in cur_indexes])
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1039 cur_limits[filter_key][cur_id] += 1
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1040 # get rid of the default dicts
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1041 final_dict = dict()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1042 for key, value_dict in cur_limits.items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1043 if len(key) == 1:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1044 key = key[0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1045 final_dict[key] = dict(value_dict)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1046 gff_handle.close()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1047 return final_dict
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1048
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1049 @_file_or_handle
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1050 def parent_child_map(self, gff_handle):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1051 """Provide a mapping of parent to child relationships in the file.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1052
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1053 Returns a dictionary of parent child relationships:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1054
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1055 keys -- tuple of (source, type) for each parent
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1056 values -- tuple of (source, type) as children of that parent
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1057
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1058 Not a parallelized map-reduce implementation.
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1059 """
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1060 # collect all of the parent and child types mapped to IDs
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1061 parent_sts = dict()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1062 child_sts = collections.defaultdict(list)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1063 for line in gff_handle:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1064 # when we hit FASTA sequences, we are done with annotations
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1065 if line.startswith("##FASTA"):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1066 break
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1067 if line.strip() and not line.startswith("#"):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1068 line_type, line_info = _gff_line_map(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1069 line, self._get_local_params()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1070 )[0]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1071 if line_type == "parent" or (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1072 line_type == "child" and line_info["id"]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1073 ):
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1074 parent_sts[line_info["id"]] = (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1075 line_info["quals"].get("source", [""])[0],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1076 line_info["type"],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1077 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1078 if line_type == "child":
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1079 for parent_id in line_info["quals"]["Parent"]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1080 child_sts[parent_id].append(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1081 (
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1082 line_info["quals"].get(
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1083 "source", [""]
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1084 )[0],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1085 line_info["type"],
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1086 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1087 )
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1088 # print parent_sts, child_sts
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1089 # generate a dictionary of the unique final type relationships
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1090 pc_map = collections.defaultdict(list)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1091 for parent_id, parent_type in parent_sts.items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1092 for child_type in child_sts[parent_id]:
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1093 pc_map[parent_type].append(child_type)
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1094 pc_final_map = dict()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1095 for ptype, ctypes in pc_map.items():
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1096 unique_ctypes = list(set(ctypes))
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1097 unique_ctypes.sort()
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1098 pc_final_map[ptype] = unique_ctypes
4c201a3d4755 planemo upload for repository https://github.com/usegalaxy-eu/temporary-tools/tree/master/jbrowse2 commit a37bfdfc108501b11c7b2aa15efb1bd16f0c4b66
fubar
parents:
diff changeset
1099 return pc_final_map