Mercurial > repos > devteam > fasta_concatenate_by_species
annotate utils/maf_utilities.py @ 3:25b8736c627a draft default tip
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit 34a6c9f94a5722bb7d2f887618aafa410a770e91"
author | devteam |
---|---|
date | Mon, 02 Mar 2020 06:47:07 -0500 |
parents | 16df616b39e5 |
children |
rev | line source |
---|---|
0 | 1 #!/usr/bin/env python |
2 """ | |
3 Provides wrappers and utilities for working with MAF files and alignments. | |
4 """ | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
5 # Dan Blankenberg |
0 | 6 import bx.align.maf |
7 import bx.intervals | |
8 import bx.interval_index_file | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
9 import sys |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
10 import os |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
11 import tempfile |
0 | 12 import logging |
13 from copy import deepcopy | |
14 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
15 try: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
16 maketrans = str.maketrans |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
17 except AttributeError: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
18 from string import maketrans |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
19 |
0 | 20 log = logging.getLogger(__name__) |
21 | |
22 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
23 GAP_CHARS = ['-'] |
0 | 24 SRC_SPLIT_CHAR = '.' |
25 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
26 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
27 def src_split(src): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
28 fields = src.split(SRC_SPLIT_CHAR, 1) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
29 spec = fields.pop(0) |
0 | 30 if fields: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
31 chrom = fields.pop(0) |
0 | 32 else: |
33 chrom = spec | |
34 return spec, chrom | |
35 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
36 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
37 def src_merge(spec, chrom, contig=None): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
38 if None in [spec, chrom]: |
0 | 39 spec = chrom = spec or chrom |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
40 return bx.align.maf.src_merge(spec, chrom, contig) |
0 | 41 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
42 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
43 def get_species_in_block(block): |
0 | 44 species = [] |
45 for c in block.components: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
46 spec, chrom = src_split(c.src) |
0 | 47 if spec not in species: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
48 species.append(spec) |
0 | 49 return species |
50 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
51 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
52 def tool_fail(msg="Unknown Error"): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
53 msg = "Fatal Error: %s" % msg |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
54 sys.exit(msg) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
55 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
56 # an object corresponding to a reference layered alignment |
0 | 57 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
58 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
59 class RegionAlignment(object): |
0 | 60 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
61 DNA_COMPLEMENT = maketrans("ACGTacgt", "TGCAtgca") |
0 | 62 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
63 def __init__(self, size, species=[]): |
0 | 64 self.size = size |
65 self.sequences = {} | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
66 if not isinstance(species, list): |
0 | 67 species = [species] |
68 for spec in species: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
69 self.add_species(spec) |
0 | 70 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
71 # add a species to the alignment |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
72 def add_species(self, species): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
73 # make temporary sequence files |
0 | 74 self.sequences[species] = tempfile.TemporaryFile() |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
75 self.sequences[species].write("-" * self.size) |
0 | 76 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
77 # returns the names for species found in alignment, skipping names as requested |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
78 def get_species_names(self, skip=[]): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
79 if not isinstance(skip, list): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
80 skip = [skip] |
0 | 81 names = self.sequences.keys() |
82 for name in skip: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
83 try: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
84 names.remove(name) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
85 except Exception: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
86 pass |
0 | 87 return names |
88 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
89 # returns the sequence for a species |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
90 def get_sequence(self, species): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
91 self.sequences[species].seek(0) |
0 | 92 return self.sequences[species].read() |
93 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
94 # returns the reverse complement of the sequence for a species |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
95 def get_sequence_reverse_complement(self, species): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
96 complement = [base for base in self.get_sequence(species).translate(self.DNA_COMPLEMENT)] |
0 | 97 complement.reverse() |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
98 return "".join(complement) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
99 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
100 # sets a position for a species |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
101 def set_position(self, index, species, base): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
102 if len(base) != 1: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
103 raise Exception("A genomic position can only have a length of 1.") |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
104 return self.set_range(index, species, base) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
105 # sets a range for a species |
0 | 106 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
107 def set_range(self, index, species, bases): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
108 if index >= self.size or index < 0: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
109 raise Exception("Your index (%i) is out of range (0 - %i)." % (index, self.size - 1)) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
110 if len(bases) == 0: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
111 raise Exception("A set of genomic positions can only have a positive length.") |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
112 if species not in self.sequences.keys(): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
113 self.add_species(species) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
114 self.sequences[species].seek(index) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
115 self.sequences[species].write(bases) |
0 | 116 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
117 # Flush temp file of specified species, or all species |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
118 def flush(self, species=None): |
0 | 119 if species is None: |
120 species = self.sequences.keys() | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
121 elif not isinstance(species, list): |
0 | 122 species = [species] |
123 for spec in species: | |
124 self.sequences[spec].flush() | |
125 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
126 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
127 class GenomicRegionAlignment(RegionAlignment): |
0 | 128 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
129 def __init__(self, start, end, species=[]): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
130 RegionAlignment.__init__(self, end - start, species) |
0 | 131 self.start = start |
132 self.end = end | |
133 | |
134 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
135 class SplicedAlignment(object): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
136 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
137 DNA_COMPLEMENT = maketrans("ACGTacgt", "TGCAtgca") |
0 | 138 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
139 def __init__(self, exon_starts, exon_ends, species=[]): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
140 if not isinstance(exon_starts, list): |
0 | 141 exon_starts = [exon_starts] |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
142 if not isinstance(exon_ends, list): |
0 | 143 exon_ends = [exon_ends] |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
144 assert len(exon_starts) == len(exon_ends), "The number of starts does not match the number of sizes." |
0 | 145 self.exons = [] |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
146 for i in range(len(exon_starts)): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
147 self.exons.append(GenomicRegionAlignment(exon_starts[i], exon_ends[i], species)) |
0 | 148 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
149 # returns the names for species found in alignment, skipping names as requested |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
150 def get_species_names(self, skip=[]): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
151 if not isinstance(skip, list): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
152 skip = [skip] |
0 | 153 names = [] |
154 for exon in self.exons: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
155 for name in exon.get_species_names(skip=skip): |
0 | 156 if name not in names: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
157 names.append(name) |
0 | 158 return names |
159 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
160 # returns the sequence for a species |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
161 def get_sequence(self, species): |
0 | 162 sequence = tempfile.TemporaryFile() |
163 for exon in self.exons: | |
164 if species in exon.get_species_names(): | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
165 sequence.write(exon.get_sequence(species)) |
0 | 166 else: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
167 sequence.write("-" * exon.size) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
168 sequence.seek(0) |
0 | 169 return sequence.read() |
170 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
171 # returns the reverse complement of the sequence for a species |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
172 def get_sequence_reverse_complement(self, species): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
173 complement = [base for base in self.get_sequence(species).translate(self.DNA_COMPLEMENT)] |
0 | 174 complement.reverse() |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
175 return "".join(complement) |
0 | 176 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
177 # Start and end of coding region |
0 | 178 @property |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
179 def start(self): |
0 | 180 return self.exons[0].start |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
181 |
0 | 182 @property |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
183 def end(self): |
0 | 184 return self.exons[-1].end |
185 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
186 # Open a MAF index using a UID |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
187 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
188 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
189 def maf_index_by_uid(maf_uid, index_location_file): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
190 for line in open(index_location_file): |
0 | 191 try: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
192 # read each line, if not enough fields, go to next line |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
193 if line[0:1] == "#": |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
194 continue |
0 | 195 fields = line.split('\t') |
196 if maf_uid == fields[1]: | |
197 try: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
198 maf_files = fields[4].replace("\n", "").replace("\r", "").split(",") |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
199 return bx.align.maf.MultiIndexed(maf_files, keep_open=True, parse_e_rows=False) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
200 except Exception as e: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
201 raise Exception('MAF UID (%s) found, but configuration appears to be malformed: %s' % (maf_uid, e)) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
202 except Exception: |
0 | 203 pass |
204 return None | |
205 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
206 # return ( index, temp_index_filename ) for user maf, if available, or build one and return it, return None when no tempfile is created |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
207 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
208 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
209 def open_or_build_maf_index(maf_file, index_filename, species=None): |
0 | 210 try: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
211 return (bx.align.maf.Indexed(maf_file, index_filename=index_filename, keep_open=True, parse_e_rows=False), None) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
212 except Exception: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
213 return build_maf_index(maf_file, species=species) |
0 | 214 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
215 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
216 def build_maf_index_species_chromosomes(filename, index_species=None): |
0 | 217 species = [] |
218 species_chromosomes = {} | |
219 indexes = bx.interval_index_file.Indexes() | |
220 blocks = 0 | |
221 try: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
222 maf_reader = bx.align.maf.Reader(open(filename)) |
0 | 223 while True: |
224 pos = maf_reader.file.tell() | |
225 block = maf_reader.next() | |
226 if block is None: | |
227 break | |
228 blocks += 1 | |
229 for c in block.components: | |
230 spec = c.src | |
231 chrom = None | |
232 if "." in spec: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
233 spec, chrom = spec.split(".", 1) |
0 | 234 if spec not in species: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
235 species.append(spec) |
0 | 236 species_chromosomes[spec] = [] |
237 if chrom and chrom not in species_chromosomes[spec]: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
238 species_chromosomes[spec].append(chrom) |
0 | 239 if index_species is None or spec in index_species: |
240 forward_strand_start = c.forward_strand_start | |
241 forward_strand_end = c.forward_strand_end | |
242 try: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
243 forward_strand_start = int(forward_strand_start) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
244 forward_strand_end = int(forward_strand_end) |
0 | 245 except ValueError: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
246 continue # start and end are not integers, can't add component to index, goto next component |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
247 # this likely only occurs when parse_e_rows is True? |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
248 # could a species exist as only e rows? should the |
0 | 249 if forward_strand_end > forward_strand_start: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
250 # require positive length; i.e. certain lines have start = end = 0 and cannot be indexed |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
251 indexes.add(c.src, forward_strand_start, forward_strand_end, pos, max=c.src_size) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
252 except Exception as e: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
253 # most likely a bad MAF |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
254 log.debug('Building MAF index on %s failed: %s' % (filename, e)) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
255 return (None, [], {}, 0) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
256 return (indexes, species, species_chromosomes, blocks) |
0 | 257 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
258 # builds and returns ( index, index_filename ) for specified maf_file |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
259 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
260 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
261 def build_maf_index(maf_file, species=None): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
262 indexes, found_species, species_chromosomes, blocks = build_maf_index_species_chromosomes(maf_file, species) |
0 | 263 if indexes is not None: |
264 fd, index_filename = tempfile.mkstemp() | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
265 out = os.fdopen(fd, 'w') |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
266 indexes.write(out) |
0 | 267 out.close() |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
268 return (bx.align.maf.Indexed(maf_file, index_filename=index_filename, keep_open=True, parse_e_rows=False), index_filename) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
269 return (None, None) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
270 |
0 | 271 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
272 def component_overlaps_region(c, region): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
273 if c is None: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
274 return False |
0 | 275 start, end = c.get_forward_strand_start(), c.get_forward_strand_end() |
276 if region.start >= end or region.end <= start: | |
277 return False | |
278 return True | |
279 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
280 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
281 def chop_block_by_region(block, src, region, species=None, mincols=0): |
0 | 282 # This chopping method was designed to maintain consistency with how start/end padding gaps have been working in Galaxy thus far: |
283 # behavior as seen when forcing blocks to be '+' relative to src sequence (ref) and using block.slice_by_component( ref, slice_start, slice_end ) | |
284 # whether-or-not this is the 'correct' behavior is questionable, but this will at least maintain consistency | |
285 # comments welcome | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
286 slice_start = block.text_size # max for the min() |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
287 slice_end = 0 # min for the max() |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
288 old_score = block.score # save old score for later use |
0 | 289 # We no longer assume only one occurance of src per block, so we need to check them all |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
290 for c in iter_components_by_src(block, src): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
291 if component_overlaps_region(c, region): |
0 | 292 if c.text is not None: |
293 rev_strand = False | |
294 if c.strand == "-": | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
295 # We want our coord_to_col coordinates to be returned from positive stranded component |
0 | 296 rev_strand = True |
297 c = c.reverse_complement() | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
298 start = max(region.start, c.start) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
299 end = min(region.end, c.end) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
300 start = c.coord_to_col(start) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
301 end = c.coord_to_col(end) |
0 | 302 if rev_strand: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
303 # need to orient slice coordinates to the original block direction |
0 | 304 slice_len = end - start |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
305 end = len(c.text) - start |
0 | 306 start = end - slice_len |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
307 slice_start = min(start, slice_start) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
308 slice_end = max(end, slice_end) |
0 | 309 |
310 if slice_start < slice_end: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
311 block = block.slice(slice_start, slice_end) |
0 | 312 if block.text_size > mincols: |
313 # restore old score, may not be accurate, but it is better than 0 for everything? | |
314 block.score = old_score | |
315 if species is not None: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
316 block = block.limit_to_species(species) |
0 | 317 block.remove_all_gap_columns() |
318 return block | |
319 return None | |
320 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
321 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
322 def orient_block_by_region(block, src, region, force_strand=None): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
323 # loop through components matching src, |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
324 # make sure each of these components overlap region |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
325 # cache strand for each of overlaping regions |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
326 # if force_strand / region.strand not in strand cache, reverse complement |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
327 # we could have 2 sequences with same src, overlapping region, on different strands, this would cause no reverse_complementing |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
328 strands = [c.strand for c in iter_components_by_src(block, src) if component_overlaps_region(c, region)] |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
329 if strands and (force_strand is None and region.strand not in strands) or (force_strand is not None and force_strand not in strands): |
0 | 330 block = block.reverse_complement() |
331 return block | |
332 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
333 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
334 def get_oriented_chopped_blocks_for_region(index, src, region, species=None, mincols=0, force_strand=None): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
335 for block, idx, offset in get_oriented_chopped_blocks_with_index_offset_for_region(index, src, region, species, mincols, force_strand): |
0 | 336 yield block |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
337 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
338 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
339 def get_oriented_chopped_blocks_with_index_offset_for_region(index, src, region, species=None, mincols=0, force_strand=None): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
340 for block, idx, offset in get_chopped_blocks_with_index_offset_for_region(index, src, region, species, mincols): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
341 yield orient_block_by_region(block, src, region, force_strand), idx, offset |
0 | 342 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
343 # split a block with multiple occurances of src into one block per src |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
344 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
345 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
346 def iter_blocks_split_by_src(block, src): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
347 for src_c in iter_components_by_src(block, src): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
348 new_block = bx.align.Alignment(score=block.score, attributes=deepcopy(block.attributes)) |
0 | 349 new_block.text_size = block.text_size |
350 for c in block.components: | |
351 if c == src_c or c.src != src: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
352 new_block.add_component(deepcopy(c)) # components have reference to alignment, dont want to loose reference to original alignment block in original components |
0 | 353 yield new_block |
354 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
355 # split a block into multiple blocks with all combinations of a species appearing only once per block |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
356 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
357 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
358 def iter_blocks_split_by_species(block, species=None): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
359 def __split_components_by_species(components_by_species, new_block): |
0 | 360 if components_by_species: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
361 # more species with components to add to this block |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
362 components_by_species = deepcopy(components_by_species) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
363 spec_comps = components_by_species.pop(0) |
0 | 364 for c in spec_comps: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
365 newer_block = deepcopy(new_block) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
366 newer_block.add_component(deepcopy(c)) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
367 for value in __split_components_by_species(components_by_species, newer_block): |
0 | 368 yield value |
369 else: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
370 # no more components to add, yield this block |
0 | 371 yield new_block |
372 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
373 # divide components by species |
0 | 374 spec_dict = {} |
375 if not species: | |
376 species = [] | |
377 for c in block.components: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
378 spec, chrom = src_split(c.src) |
0 | 379 if spec not in spec_dict: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
380 spec_dict[spec] = [] |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
381 species.append(spec) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
382 spec_dict[spec].append(c) |
0 | 383 else: |
384 for spec in species: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
385 spec_dict[spec] = [] |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
386 for c in iter_components_by_src_start(block, spec): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
387 spec_dict[spec].append(c) |
0 | 388 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
389 empty_block = bx.align.Alignment(score=block.score, attributes=deepcopy(block.attributes)) # should we copy attributes? |
0 | 390 empty_block.text_size = block.text_size |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
391 # call recursive function to split into each combo of spec/blocks |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
392 for value in __split_components_by_species(spec_dict.values(), empty_block): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
393 sort_block_components_by_block(value, block) # restore original component order |
0 | 394 yield value |
395 | |
396 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
397 # generator yielding only chopped and valid blocks for a specified region |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
398 def get_chopped_blocks_for_region(index, src, region, species=None, mincols=0): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
399 for block, idx, offset in get_chopped_blocks_with_index_offset_for_region(index, src, region, species, mincols): |
0 | 400 yield block |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
401 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
402 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
403 def get_chopped_blocks_with_index_offset_for_region(index, src, region, species=None, mincols=0): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
404 for block, idx, offset in index.get_as_iterator_with_index_and_offset(src, region.start, region.end): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
405 block = chop_block_by_region(block, src, region, species, mincols) |
0 | 406 if block is not None: |
407 yield block, idx, offset | |
408 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
409 # returns a filled region alignment for specified regions |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
410 |
0 | 411 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
412 def get_region_alignment(index, primary_species, chrom, start, end, strand='+', species=None, mincols=0, overwrite_with_gaps=True): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
413 if species is not None: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
414 alignment = RegionAlignment(end - start, species) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
415 else: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
416 alignment = RegionAlignment(end - start, primary_species) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
417 return fill_region_alignment(alignment, index, primary_species, chrom, start, end, strand, species, mincols, overwrite_with_gaps) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
418 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
419 # reduces a block to only positions exisiting in the src provided |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
420 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
421 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
422 def reduce_block_by_primary_genome(block, species, chromosome, region_start): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
423 # returns ( startIndex, {species:texts} |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
424 # where texts' contents are reduced to only positions existing in the primary genome |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
425 src = "%s.%s" % (species, chromosome) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
426 ref = block.get_component_by_src(src) |
0 | 427 start_offset = ref.start - region_start |
428 species_texts = {} | |
429 for c in block.components: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
430 species_texts[c.src.split('.')[0]] = list(c.text) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
431 # remove locations which are gaps in the primary species, starting from the downstream end |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
432 for i in range(len(species_texts[species]) - 1, -1, -1): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
433 if species_texts[species][i] == '-': |
0 | 434 for text in species_texts.values(): |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
435 text.pop(i) |
0 | 436 for spec, text in species_texts.items(): |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
437 species_texts[spec] = ''.join(text) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
438 return (start_offset, species_texts) |
0 | 439 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
440 # fills a region alignment |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
441 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
442 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
443 def fill_region_alignment(alignment, index, primary_species, chrom, start, end, strand='+', species=None, mincols=0, overwrite_with_gaps=True): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
444 region = bx.intervals.Interval(start, end) |
0 | 445 region.chrom = chrom |
446 region.strand = strand | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
447 primary_src = "%s.%s" % (primary_species, chrom) |
0 | 448 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
449 # Order blocks overlaping this position by score, lowest first |
0 | 450 blocks = [] |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
451 for block, idx, offset in index.get_as_iterator_with_index_and_offset(primary_src, start, end): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
452 score = float(block.score) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
453 for i in range(0, len(blocks)): |
0 | 454 if score < blocks[i][0]: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
455 blocks.insert(i, (score, idx, offset)) |
0 | 456 break |
457 else: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
458 blocks.append((score, idx, offset)) |
0 | 459 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
460 # gap_chars_tuple = tuple( GAP_CHARS ) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
461 gap_chars_str = ''.join(GAP_CHARS) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
462 # Loop through ordered blocks and layer by increasing score |
0 | 463 for block_dict in blocks: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
464 for block in iter_blocks_split_by_species(block_dict[1].get_at_offset(block_dict[2])): # need to handle each occurance of sequence in block seperately |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
465 if component_overlaps_region(block.get_component_by_src(primary_src), region): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
466 block = chop_block_by_region(block, primary_src, region, species, mincols) # chop block |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
467 block = orient_block_by_region(block, primary_src, region) # orient block |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
468 start_offset, species_texts = reduce_block_by_primary_genome(block, primary_species, chrom, start) |
0 | 469 for spec, text in species_texts.items(): |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
470 # we should trim gaps from both sides, since these are not positions in this species genome (sequence) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
471 text = text.rstrip(gap_chars_str) |
0 | 472 gap_offset = 0 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
473 while True in [text.startswith(gap_char) for gap_char in GAP_CHARS]: # python2.4 doesn't accept a tuple for .startswith() |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
474 # while text.startswith( gap_chars_tuple ): |
0 | 475 gap_offset += 1 |
476 text = text[1:] | |
477 if not text: | |
478 break | |
479 if text: | |
480 if overwrite_with_gaps: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
481 alignment.set_range(start_offset + gap_offset, spec, text) |
0 | 482 else: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
483 for i, char in enumerate(text): |
0 | 484 if char not in GAP_CHARS: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
485 alignment.set_position(start_offset + gap_offset + i, spec, char) |
0 | 486 return alignment |
487 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
488 # returns a filled spliced region alignment for specified region with start and end lists |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
489 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
490 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
491 def get_spliced_region_alignment(index, primary_species, chrom, starts, ends, strand='+', species=None, mincols=0, overwrite_with_gaps=True): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
492 # create spliced alignment object |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
493 if species is not None: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
494 alignment = SplicedAlignment(starts, ends, species) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
495 else: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
496 alignment = SplicedAlignment(starts, ends, [primary_species]) |
0 | 497 for exon in alignment.exons: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
498 fill_region_alignment(exon, index, primary_species, chrom, exon.start, exon.end, strand, species, mincols, overwrite_with_gaps) |
0 | 499 return alignment |
500 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
501 # loop through string array, only return non-commented lines |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
502 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
503 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
504 def line_enumerator(lines, comment_start='#'): |
0 | 505 i = 0 |
506 for line in lines: | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
507 if not line.startswith(comment_start): |
0 | 508 i += 1 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
509 yield (i, line) |
0 | 510 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
511 # read a GeneBed file, return list of starts, ends, raw fields |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
512 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
513 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
514 def get_starts_ends_fields_from_gene_bed(line): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
515 # Starts and ends for exons |
0 | 516 starts = [] |
517 ends = [] | |
518 | |
519 fields = line.split() | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
520 # Requires atleast 12 BED columns |
0 | 521 if len(fields) < 12: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
522 raise Exception("Not a proper 12 column BED line (%s)." % line) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
523 tx_start = int(fields[1]) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
524 strand = fields[5] |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
525 if strand != '-': |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
526 strand = '+' # Default strand is + |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
527 cds_start = int(fields[6]) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
528 cds_end = int(fields[7]) |
0 | 529 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
530 # Calculate and store starts and ends of coding exons |
0 | 531 region_start, region_end = cds_start, cds_end |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
532 exon_starts = map(int, fields[11].rstrip(',\n').split(',')) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
533 exon_starts = map((lambda x: x + tx_start), exon_starts) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
534 exon_ends = map(int, fields[10].rstrip(',').split(',')) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
535 exon_ends = map((lambda x, y: x + y), exon_starts, exon_ends) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
536 for start, end in zip(exon_starts, exon_ends): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
537 start = max(start, region_start) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
538 end = min(end, region_end) |
0 | 539 if start < end: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
540 starts.append(start) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
541 ends.append(end) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
542 return (starts, ends, fields) |
0 | 543 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
544 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
545 def iter_components_by_src(block, src): |
0 | 546 for c in block.components: |
547 if c.src == src: | |
548 yield c | |
549 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
550 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
551 def get_components_by_src(block, src): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
552 return [value for value in iter_components_by_src(block, src)] |
0 | 553 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
554 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
555 def iter_components_by_src_start(block, src): |
0 | 556 for c in block.components: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
557 if c.src.startswith(src): |
0 | 558 yield c |
559 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
560 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
561 def get_components_by_src_start(block, src): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
562 return [value for value in iter_components_by_src_start(block, src)] |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
563 |
0 | 564 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
565 def sort_block_components_by_block(block1, block2): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
566 # orders the components in block1 by the index of the component in block2 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
567 # block1 must be a subset of block2 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
568 # occurs in-place |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
569 return block1.components.sort(cmp=lambda x, y: block2.components.index(x) - block2.components.index(y)) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
570 |
0 | 571 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
572 def get_species_in_maf(maf_filename): |
0 | 573 species = [] |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
574 for block in bx.align.maf.Reader(open(maf_filename)): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
575 for spec in get_species_in_block(block): |
0 | 576 if spec not in species: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
577 species.append(spec) |
0 | 578 return species |
579 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
580 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
581 def parse_species_option(species): |
0 | 582 if species: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
583 species = species.split(',') |
0 | 584 if 'None' not in species: |
585 return species | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
586 return None # provided species was '', None, or had 'None' in it |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
587 |
0 | 588 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
589 def remove_temp_index_file(index_filename): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
590 try: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
591 os.unlink(index_filename) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
592 except Exception: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
593 pass |
0 | 594 |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
595 # Below are methods to deal with FASTA files |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
596 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
597 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
598 def get_fasta_header(component, attributes={}, suffix=None): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
599 header = ">%s(%s):%i-%i|" % (component.src, component.strand, component.get_forward_strand_start(), component.get_forward_strand_end()) |
0 | 600 for key, value in attributes.iteritems(): |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
601 header = "%s%s=%s|" % (header, key, value) |
0 | 602 if suffix: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
603 header = "%s%s" % (header, suffix) |
0 | 604 else: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
605 header = "%s%s" % (header, src_split(component.src)[0]) |
0 | 606 return header |
607 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
608 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
609 def get_attributes_from_fasta_header(header): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
610 if not header: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
611 return {} |
0 | 612 attributes = {} |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
613 header = header.lstrip('>') |
0 | 614 header = header.strip() |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
615 fields = header.split('|') |
0 | 616 try: |
617 region = fields[0] | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
618 region = region.split('(', 1) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
619 temp = region[0].split('.', 1) |
0 | 620 attributes['species'] = temp[0] |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
621 if len(temp) == 2: |
0 | 622 attributes['chrom'] = temp[1] |
623 else: | |
624 attributes['chrom'] = temp[0] | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
625 region = region[1].split(')', 1) |
0 | 626 attributes['strand'] = region[0] |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
627 region = region[1].lstrip(':').split('-') |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
628 attributes['start'] = int(region[0]) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
629 attributes['end'] = int(region[1]) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
630 except Exception: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
631 # fields 0 is not a region coordinate |
0 | 632 pass |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
633 if len(fields) > 2: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
634 for i in range(1, len(fields) - 1): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
635 prop = fields[i].split('=', 1) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
636 if len(prop) == 2: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
637 attributes[prop[0]] = prop[1] |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
638 if len(fields) > 1: |
0 | 639 attributes['__suffix__'] = fields[-1] |
640 return attributes | |
641 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
642 |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
643 def iter_fasta_alignment(filename): |
0 | 644 class fastaComponent: |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
645 def __init__(self, species, text=""): |
0 | 646 self.species = species |
647 self.text = text | |
648 | |
2
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
649 def extend(self, text): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
650 self.text = self.text + text.replace('\n', '').replace('\r', '').strip() |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
651 # yields a list of fastaComponents for a FASTA file |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
652 with open(filename, 'r') as f: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
653 components = [] |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
654 # cur_component = None |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
655 while True: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
656 line = f.readline() |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
657 if not line: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
658 if components: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
659 yield components |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
660 return |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
661 line = line.strip() |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
662 if not line: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
663 if components: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
664 yield components |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
665 components = [] |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
666 elif line.startswith('>'): |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
667 attributes = get_attributes_from_fasta_header(line) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
668 components.append(fastaComponent(attributes['species'])) |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
669 elif components: |
16df616b39e5
"planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/fasta_concatenate_by_species commit cd1ed08574b749eee2a3f6e6151dbb0c8ca15bbf"
devteam
parents:
1
diff
changeset
|
670 components[-1].extend(line) |