Mercurial > repos > earlhaminst > ete

annotate ete_homology_classifier.py @ 5:817031b8486d draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
author earlhaminst
date Thu, 10 May 2018 06:15:17 -0400
parents
children f1eca1158f21
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
1 from __future__ import print_function
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
2
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
3 import optparse
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
4
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
5 from ete3 import PhyloTree
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
6
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
7
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
8 def main():
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
9 usage = "usage: %prog --genetree <genetree-file> --speciestree <speciestree-file> [options]"
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
10 parser = optparse.OptionParser(usage=usage)
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
11 parser.add_option('--genetree', help='GeneTree in nhx format')
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
12 parser.add_option('--out_format', type='string', default='tabular', help='Choose output format')
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
13 parser.add_option('--filters', default='', help='Filter families')
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
14
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
15 options, args = parser.parse_args()
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
16
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
17 if options.genetree is None:
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
18 parser.error("--genetree option must be specified, GeneTree in nhx format")
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
19
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
20 # reads single gene tree
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
21 genetree = PhyloTree(options.genetree)
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
22
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
23 leaves_list = genetree.get_leaf_names()
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
24 # Genetree nodes are required to be in gene_species format
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
25 leaves_list = [_ for _ in leaves_list if '_' in _]
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
26
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
27 species_list = [_.split("_")[1] for _ in leaves_list]
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
28
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
29 species_dict = {}
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
30 for species in species_list:
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
31 count = "one"
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
32 if species in species_dict:
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
33 count = "many"
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
34 species_dict[species] = count
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
35
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
36 homologies = {
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
37 'one-to-one': [],
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
38 'one-to-many': [],
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
39 'many-to-one': [],
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
40 'many-to-many': [],
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
41 'paralogs': []
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
42 }
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
43
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
44 # stores relevant homology types in dict
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
45 for i, leaf1 in enumerate(leaves_list):
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
46 for leaf2 in leaves_list[i + 1:]:
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
47 id1 = leaf1.split(":")[1] if ":" in leaf1 else leaf1
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
48 id2 = leaf2.split(":")[1] if ":" in leaf2 else leaf2
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
49 species1 = id1.split("_")[1]
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
50 species2 = id2.split("_")[1]
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
51 if species1 == species2:
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
52 homology_type = 'paralogs'
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
53 else:
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
54 homology_type = species_dict[species1] + "-to-" + species_dict[species2]
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
55 homologies[homology_type].append((id1, id2))
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
56
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
57 options.filters = options.filters.split(",")
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
58
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
59 if options.out_format == 'tabular':
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
60 for homology_type, homologs_list in homologies.items():
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
61 # checks if homology type is in filter
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
62 if homology_type in options.filters:
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
63 for (gene1, gene2) in homologs_list:
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
64 print("%s\t%s\t%s" % (gene1, gene2, homology_type))
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
65 elif options.out_format == 'csv':
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
66 print_family = True
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
67 for homology_type, homologs_list in homologies.items():
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
68 if homologs_list and homology_type not in options.filters:
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
69 print_family = False
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
70 break
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
71
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
72 # prints family if homology type is not found in filter
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
73 if print_family:
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
74 print(','.join(leaves_list))
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
75
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
76
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
77 if __name__ == "__main__":
817031b8486d planemo upload for repository https://github.com/TGAC/earlham-galaxytools/tree/master/tools/ete commit b97aee603b9acf29981719160e963a1efe2946d0
earlhaminst
parents:
diff changeset
78 main()