Mercurial > repos > bebatut > compare_humann2_output

annotate compare_humann2_output.py @ 3:eaa95ea1195c draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
author bgruening
date Wed, 19 Oct 2022 14:49:06 +0000
parents 05766022dfc4
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
1 #!/usr/bin/env python
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
2 # -*- coding: utf-8 -*-
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
3
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
4 import argparse
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
5
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
6
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
7 def extract_abundances(fp, nb_charact_to_extract):
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
8 abundances = {}
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
9 more_abund_charact = []
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
10 abund_sum = 0
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
11 with open(fp, "r") as abundance_f:
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
12 for line in abundance_f.readlines()[1:]:
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
13 split_line = line[:-1].split("\t")
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
14 charact_id = split_line[0]
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
15 abund = float(split_line[1])
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
16 abundances[charact_id] = 100 * abund
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
17 abund_sum += abundances[charact_id]
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
18
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
19 if len(more_abund_charact) < nb_charact_to_extract:
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
20 more_abund_charact.append(charact_id)
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
21 else:
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
22 best_pos = None
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
23 for i in range(len(more_abund_charact) - 1, -1, -1):
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
24 if abundances[more_abund_charact[i]] < abund:
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
25 best_pos = i
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
26 else:
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
27 break
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
28 if best_pos is not None:
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
29 tmp_more_abund_charact = more_abund_charact
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
30 more_abund_charact = tmp_more_abund_charact[:best_pos]
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
31 more_abund_charact += [charact_id]
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
32 more_abund_charact += tmp_more_abund_charact[best_pos:-1]
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
33 return abundances, more_abund_charact
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
34
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
35
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
36 def format_characteristic_name(all_name):
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
37 if all_name.find(":") != -1:
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
38 charact_id = all_name.split(":")[0]
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
39 char_name = all_name.split(":")[1][1:]
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
40 else:
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
41 charact_id = all_name
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
42 char_name = ""
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
43
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
44 char_name = char_name.replace("/", " ")
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
45 char_name = char_name.replace("-", " ")
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
46 char_name = char_name.replace("'", "")
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
47 if char_name.find("(") != -1 and char_name.find(")") != -1:
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
48 open_bracket = char_name.find("(")
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
49 close_bracket = char_name.find(")") + 1
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
50 char_name = char_name[:open_bracket] + char_name[close_bracket:]
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
51 return charact_id, char_name
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
52
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
53
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
54 def write_more_abundant_charat(abundances, more_abund_charact, output_fp):
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
55 with open(output_fp, "w") as output_f:
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
56 output_f.write("id\tname\t%s\n" % "\t".join(abundances.keys()))
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
57
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
58 for mac in more_abund_charact:
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
59 charact_id, charact_name = format_characteristic_name(mac)
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
60 output_f.write("%s\t%s" % (charact_id, charact_name))
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
61 for sample in abundances:
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
62 abund = abundances[sample].get(mac, 0)
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
63 output_f.write("\t%s" % (abund))
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
64 output_f.write("\n")
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
65
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
66
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
67 def extract_similar_characteristics(abund, sim_output_fp, output_files):
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
68 abund_keys = list(abund)
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
69 sim_characteristics = set(abund[abund_keys[0]].keys())
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
70 for sample in abund_keys[1:]:
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
71 sim_characteristics.intersection_update(abund[sample].keys())
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
72 print("Similar between all samples: %s" % len(sim_characteristics))
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
73
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
74 with open(sim_output_fp, "w") as sim_output_f:
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
75 sim_output_f.write("id\tname\t%s\n" % "\t".join(abund_keys))
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
76 for charact in list(sim_characteristics):
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
77 charact_id, charact_name = format_characteristic_name(charact)
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
78 sim_output_f.write("%s\t%s" % (charact_id, charact_name))
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
79 for sample in abund_keys:
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
80 sim_output_f.write("\t%s" % abund[sample][charact])
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
81 sim_output_f.write("\n")
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
82
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
83 print("Specific to samples:")
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
84 diff_char = {}
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
85 for i in range(len(abund_keys)):
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
86 sample = abund_keys[i]
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
87 print(" %s" % sample)
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
88 print(" All: %s" % len(abund[sample].keys()))
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
89 diff_char[sample] = set(abund[sample].keys())
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
90 diff_char[sample].difference_update(sim_characteristics)
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
91 perc = 100 * len(diff_char[sample]) / (1.0 * len(abund[sample].keys()))
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
92 print(" Number of specific characteristics: %s" % len(diff_char[sample]))
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
93 print(" Percentage of specific characteristics: %s" % perc)
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
94
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
95 relative_abundance = 0
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
96 with open(output_files[i], "w") as output_f:
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
97 output_f.write("id\tname\tabundances\n")
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
98 for charact in list(diff_char[sample]):
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
99 charact_id, charact_name = format_characteristic_name(charact)
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
100 output_f.write("%s\t%s" % (charact_id, charact_name))
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
101 output_f.write("%s\n" % abund[sample][charact])
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
102 relative_abundance += abund[sample][charact]
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
103 print(
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
104 " Relative abundance of specific characteristics: %s"
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
105 % relative_abundance
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
106 )
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
107
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
108 return sim_characteristics
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
109
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
110
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
111 def compare_humann2_output(args):
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
112 abund = {}
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
113 more_abund_charact = []
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
114
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
115 for i in range(len(args.sample_name)):
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
116 abund[args.sample_name[i]], mac = extract_abundances(
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
117 args.charact_input_fp[i], args.most_abundant_characteristics_to_extract
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
118 )
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
119 more_abund_charact += mac
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
120
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
121 write_more_abundant_charat(
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
122 abund, list(set(more_abund_charact)), args.more_abundant_output_fp
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
123 )
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
124 extract_similar_characteristics(
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
125 abund, args.similar_output_fp, args.specific_output_fp
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
126 )
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
127
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
128
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
129 if __name__ == "__main__":
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
130 parser = argparse.ArgumentParser()
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
131 parser.add_argument("--sample_name", required=True, action="append")
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
132 parser.add_argument("--charact_input_fp", required=True, action="append")
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
133 parser.add_argument(
3
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
134 "--most_abundant_characteristics_to_extract", required=True, type=int
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
135 )
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
136 parser.add_argument("--more_abundant_output_fp", required=True)
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
137 parser.add_argument("--similar_output_fp", required=True)
eaa95ea1195c "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/compare_humann2_output commit eea46077010e699403ce6995d7d4aac77b2e0b43"
bgruening
parents: 2
diff changeset
138 parser.add_argument("--specific_output_fp", required=True, action="append")
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
139 args = parser.parse_args()
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
140
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
141 if len(args.sample_name) != len(args.charact_input_fp):
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
142 string = "Same number of values (in same order) are expected for "
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
143 string += "--sample_name and --charact_input_fp"
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
144 raise ValueError(string)
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
145 if len(args.sample_name) != len(args.specific_output_fp):
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
146 string = "Same number of values (in same order) are expected for "
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
147 string += "--sample_name and --specific_output_fp"
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
148 raise ValueError(string)
0
9959fa526f1a planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit c16428041ae3d60b61b6570035c9268726730543-dirty
bebatut
parents:
diff changeset
149
2
05766022dfc4 "planemo upload for repository https://github.com/asaim/galaxytools/tree/master/tools/compare_humann2_output commit dc55dc3b5275d1d6aac390698c0c6e0ab8fbf2f7"
bebatut
parents: 0
diff changeset
150 compare_humann2_output(args)