miga-base 0.3.0.5 → 0.3.0.6
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/lib/miga/dataset_result.rb +4 -1
- data/lib/miga/version.rb +2 -2
- data/utils/enveomics/Pipelines/assembly.pbs/FastA.N50.pl +1 -0
- data/utils/enveomics/Pipelines/assembly.pbs/FastA.filterN.pl +1 -0
- data/utils/enveomics/Pipelines/assembly.pbs/FastA.length.pl +1 -0
- data/utils/enveomics/Pipelines/blast.pbs/FastA.split.pl +1 -0
- data/utils/enveomics/Scripts/lib/enveomics.R +1 -0
- metadata +165 -158
- data/utils/enveomics/Pipelines/assembly.pbs/FastA.N50.pl +0 -56
- data/utils/enveomics/Pipelines/assembly.pbs/FastA.filterN.pl +0 -60
- data/utils/enveomics/Pipelines/assembly.pbs/FastA.length.pl +0 -38
- data/utils/enveomics/Pipelines/blast.pbs/FastA.split.pl +0 -55
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA1:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: e86b0a3fd7e0ef5b3d2ac6c82e0d6e44b595dfa7
|
|
4
|
+
data.tar.gz: d0c718124aaee78e75f52db4679a29958e7df193
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: a1ff8aee53f8d3c2c10124cb6de6d363b6c8d15e842b1454b025ffc7a90580c4b06db573b0fc0290b802d4d0ba1344d952f41aa266141303f4dbb4502d20368e
|
|
7
|
+
data.tar.gz: d84243fe1316b9d475fa43fe7559a712a5533b012d69268b1e92e4e47e6f75ed0c603dd07b3e2350fa9040c8188c1524250a92677a2bf534b2425602d9d6b6ba
|
data/lib/miga/dataset_result.rb
CHANGED
|
@@ -148,7 +148,10 @@ module MiGA::DatasetResult
|
|
|
148
148
|
result_files_exist?(base, ".nomytaxa.txt")
|
|
149
149
|
r = MiGA::Result.new("#{base}.json")
|
|
150
150
|
add_files_to_ds_result(r, name, mytaxa:".mytaxa", blast:".blast",
|
|
151
|
-
mytaxain:".mytaxain", nomytaxa:".nomytaxa.txt"
|
|
151
|
+
mytaxain:".mytaxain", nomytaxa:".nomytaxa.txt",
|
|
152
|
+
species:".mytaxa.Species.txt", genus:".mytaxa.Genus.txt",
|
|
153
|
+
phylum:".mytaxa.Phylum.txt", innominate:".mytaxa.innominate",
|
|
154
|
+
kronain:".mytaxa.krona", krona:".html")
|
|
152
155
|
else
|
|
153
156
|
MiGA::Result.new("#{base}.json")
|
|
154
157
|
end
|
data/lib/miga/version.rb
CHANGED
|
@@ -10,7 +10,7 @@ module MiGA
|
|
|
10
10
|
# - Float representing the major.minor version.
|
|
11
11
|
# - Integer representing gem releases of the current version.
|
|
12
12
|
# - Integer representing minor changes that require new version number.
|
|
13
|
-
VERSION = [0.3, 0,
|
|
13
|
+
VERSION = [0.3, 0, 6]
|
|
14
14
|
|
|
15
15
|
##
|
|
16
16
|
# Nickname for the current major.minor version.
|
|
@@ -18,7 +18,7 @@ module MiGA
|
|
|
18
18
|
|
|
19
19
|
##
|
|
20
20
|
# Date of the current gem release.
|
|
21
|
-
VERSION_DATE = Date.new(2017,
|
|
21
|
+
VERSION_DATE = Date.new(2017, 8, 24)
|
|
22
22
|
|
|
23
23
|
##
|
|
24
24
|
# Reference of MiGA.
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
utils/enveomics/Pipelines/assembly.pbs/../../Scripts/FastA.N50.pl
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
utils/enveomics/Pipelines/assembly.pbs/../../Scripts/FastA.filterN.pl
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
utils/enveomics/Pipelines/assembly.pbs/../../Scripts/FastA.length.pl
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
utils/enveomics/Pipelines/blast.pbs/../../Scripts/FastA.split.pl
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
utils/enveomics/Scripts/lib/../../enveomics.R
|
metadata
CHANGED
|
@@ -1,103 +1,109 @@
|
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
|
2
2
|
name: miga-base
|
|
3
3
|
version: !ruby/object:Gem::Version
|
|
4
|
-
version: 0.3.0.
|
|
4
|
+
version: 0.3.0.6
|
|
5
5
|
platform: ruby
|
|
6
6
|
authors:
|
|
7
7
|
- Luis M. Rodriguez-R
|
|
8
8
|
autorequire:
|
|
9
9
|
bindir: bin
|
|
10
10
|
cert_chain: []
|
|
11
|
-
date: 2017-
|
|
11
|
+
date: 2017-08-24 00:00:00.000000000 Z
|
|
12
12
|
dependencies:
|
|
13
13
|
- !ruby/object:Gem::Dependency
|
|
14
14
|
name: rest-client
|
|
15
15
|
requirement: !ruby/object:Gem::Requirement
|
|
16
16
|
requirements:
|
|
17
|
-
- - ~>
|
|
17
|
+
- - "~>"
|
|
18
18
|
- !ruby/object:Gem::Version
|
|
19
19
|
version: '1.7'
|
|
20
20
|
type: :runtime
|
|
21
21
|
prerelease: false
|
|
22
22
|
version_requirements: !ruby/object:Gem::Requirement
|
|
23
23
|
requirements:
|
|
24
|
-
- - ~>
|
|
24
|
+
- - "~>"
|
|
25
25
|
- !ruby/object:Gem::Version
|
|
26
26
|
version: '1.7'
|
|
27
27
|
- !ruby/object:Gem::Dependency
|
|
28
28
|
name: daemons
|
|
29
29
|
requirement: !ruby/object:Gem::Requirement
|
|
30
30
|
requirements:
|
|
31
|
-
- - ~>
|
|
31
|
+
- - "~>"
|
|
32
32
|
- !ruby/object:Gem::Version
|
|
33
33
|
version: '1.2'
|
|
34
|
-
- -
|
|
34
|
+
- - ">="
|
|
35
35
|
- !ruby/object:Gem::Version
|
|
36
36
|
version: 1.2.4
|
|
37
37
|
type: :runtime
|
|
38
38
|
prerelease: false
|
|
39
39
|
version_requirements: !ruby/object:Gem::Requirement
|
|
40
40
|
requirements:
|
|
41
|
-
- - ~>
|
|
41
|
+
- - "~>"
|
|
42
42
|
- !ruby/object:Gem::Version
|
|
43
43
|
version: '1.2'
|
|
44
|
-
- -
|
|
44
|
+
- - ">="
|
|
45
45
|
- !ruby/object:Gem::Version
|
|
46
46
|
version: 1.2.4
|
|
47
47
|
- !ruby/object:Gem::Dependency
|
|
48
48
|
name: json
|
|
49
49
|
requirement: !ruby/object:Gem::Requirement
|
|
50
50
|
requirements:
|
|
51
|
-
- -
|
|
51
|
+
- - ">"
|
|
52
52
|
- !ruby/object:Gem::Version
|
|
53
53
|
version: '1.8'
|
|
54
|
+
- - "<"
|
|
55
|
+
- !ruby/object:Gem::Version
|
|
56
|
+
version: '3'
|
|
54
57
|
type: :runtime
|
|
55
58
|
prerelease: false
|
|
56
59
|
version_requirements: !ruby/object:Gem::Requirement
|
|
57
60
|
requirements:
|
|
58
|
-
- -
|
|
61
|
+
- - ">"
|
|
59
62
|
- !ruby/object:Gem::Version
|
|
60
63
|
version: '1.8'
|
|
64
|
+
- - "<"
|
|
65
|
+
- !ruby/object:Gem::Version
|
|
66
|
+
version: '3'
|
|
61
67
|
- !ruby/object:Gem::Dependency
|
|
62
68
|
name: sqlite3
|
|
63
69
|
requirement: !ruby/object:Gem::Requirement
|
|
64
70
|
requirements:
|
|
65
|
-
- - ~>
|
|
71
|
+
- - "~>"
|
|
66
72
|
- !ruby/object:Gem::Version
|
|
67
73
|
version: '1.3'
|
|
68
74
|
type: :runtime
|
|
69
75
|
prerelease: false
|
|
70
76
|
version_requirements: !ruby/object:Gem::Requirement
|
|
71
77
|
requirements:
|
|
72
|
-
- - ~>
|
|
78
|
+
- - "~>"
|
|
73
79
|
- !ruby/object:Gem::Version
|
|
74
80
|
version: '1.3'
|
|
75
81
|
- !ruby/object:Gem::Dependency
|
|
76
82
|
name: rake
|
|
77
83
|
requirement: !ruby/object:Gem::Requirement
|
|
78
84
|
requirements:
|
|
79
|
-
- - ~>
|
|
85
|
+
- - "~>"
|
|
80
86
|
- !ruby/object:Gem::Version
|
|
81
87
|
version: '11'
|
|
82
88
|
type: :development
|
|
83
89
|
prerelease: false
|
|
84
90
|
version_requirements: !ruby/object:Gem::Requirement
|
|
85
91
|
requirements:
|
|
86
|
-
- - ~>
|
|
92
|
+
- - "~>"
|
|
87
93
|
- !ruby/object:Gem::Version
|
|
88
94
|
version: '11'
|
|
89
95
|
- !ruby/object:Gem::Dependency
|
|
90
96
|
name: test-unit
|
|
91
97
|
requirement: !ruby/object:Gem::Requirement
|
|
92
98
|
requirements:
|
|
93
|
-
- - ~>
|
|
99
|
+
- - "~>"
|
|
94
100
|
- !ruby/object:Gem::Version
|
|
95
101
|
version: '3'
|
|
96
102
|
type: :development
|
|
97
103
|
prerelease: false
|
|
98
104
|
version_requirements: !ruby/object:Gem::Requirement
|
|
99
105
|
requirements:
|
|
100
|
-
- - ~>
|
|
106
|
+
- - "~>"
|
|
101
107
|
- !ruby/object:Gem::Version
|
|
102
108
|
version: '3'
|
|
103
109
|
description: Microbial Genomes Atlas
|
|
@@ -108,6 +114,34 @@ extensions: []
|
|
|
108
114
|
extra_rdoc_files:
|
|
109
115
|
- README.md
|
|
110
116
|
files:
|
|
117
|
+
- Gemfile
|
|
118
|
+
- LICENSE
|
|
119
|
+
- README.md
|
|
120
|
+
- Rakefile
|
|
121
|
+
- actions/about.rb
|
|
122
|
+
- actions/add.rb
|
|
123
|
+
- actions/add_result.rb
|
|
124
|
+
- actions/daemon.rb
|
|
125
|
+
- actions/date.rb
|
|
126
|
+
- actions/files.rb
|
|
127
|
+
- actions/find.rb
|
|
128
|
+
- actions/get.rb
|
|
129
|
+
- actions/init.rb
|
|
130
|
+
- actions/ln.rb
|
|
131
|
+
- actions/ls.rb
|
|
132
|
+
- actions/new.rb
|
|
133
|
+
- actions/plugins.rb
|
|
134
|
+
- actions/rm.rb
|
|
135
|
+
- actions/run.rb
|
|
136
|
+
- actions/stats.rb
|
|
137
|
+
- actions/tax_dist.rb
|
|
138
|
+
- actions/tax_index.rb
|
|
139
|
+
- actions/tax_set.rb
|
|
140
|
+
- actions/tax_test.rb
|
|
141
|
+
- bin/miga
|
|
142
|
+
- lib/miga.rb
|
|
143
|
+
- lib/miga/_data/aai-intax.tsv.gz
|
|
144
|
+
- lib/miga/_data/aai-novel.tsv.gz
|
|
111
145
|
- lib/miga/common.rb
|
|
112
146
|
- lib/miga/daemon.rb
|
|
113
147
|
- lib/miga/dataset.rb
|
|
@@ -121,18 +155,6 @@ files:
|
|
|
121
155
|
- lib/miga/tax_index.rb
|
|
122
156
|
- lib/miga/taxonomy.rb
|
|
123
157
|
- lib/miga/version.rb
|
|
124
|
-
- lib/miga.rb
|
|
125
|
-
- test/common_test.rb
|
|
126
|
-
- test/daemon_test.rb
|
|
127
|
-
- test/dataset_test.rb
|
|
128
|
-
- test/metadata_test.rb
|
|
129
|
-
- test/project_test.rb
|
|
130
|
-
- test/remote_dataset_test.rb
|
|
131
|
-
- test/tax_index_test.rb
|
|
132
|
-
- test/taxonomy_test.rb
|
|
133
|
-
- test/test_helper.rb
|
|
134
|
-
- lib/miga/_data/aai-intax.tsv.gz
|
|
135
|
-
- lib/miga/_data/aai-novel.tsv.gz
|
|
136
158
|
- scripts/_distances_functions.bash
|
|
137
159
|
- scripts/_distances_noref_nomulti.bash
|
|
138
160
|
- scripts/_distances_ref_nomulti.bash
|
|
@@ -156,74 +178,23 @@ files:
|
|
|
156
178
|
- scripts/subclades.bash
|
|
157
179
|
- scripts/trimmed_fasta.bash
|
|
158
180
|
- scripts/trimmed_reads.bash
|
|
181
|
+
- test/common_test.rb
|
|
182
|
+
- test/daemon_test.rb
|
|
183
|
+
- test/dataset_test.rb
|
|
184
|
+
- test/metadata_test.rb
|
|
185
|
+
- test/project_test.rb
|
|
186
|
+
- test/remote_dataset_test.rb
|
|
187
|
+
- test/tax_index_test.rb
|
|
188
|
+
- test/taxonomy_test.rb
|
|
189
|
+
- test/test_helper.rb
|
|
159
190
|
- utils/adapters.fa
|
|
160
191
|
- utils/core-pan-plot.R
|
|
161
|
-
- utils/enveomics/build_enveomics_r.bash
|
|
162
|
-
- utils/enveomics/enveomics.R/data/growth.curves.rda
|
|
163
|
-
- utils/enveomics/enveomics.R/data/phyla.counts.rda
|
|
164
|
-
- utils/enveomics/enveomics.R/DESCRIPTION
|
|
165
|
-
- utils/enveomics/enveomics.R/man/enve.__prune.iter.Rd
|
|
166
|
-
- utils/enveomics/enveomics.R/man/enve.__prune.reduce.Rd
|
|
167
|
-
- utils/enveomics/enveomics.R/man/enve.__tribs.Rd
|
|
168
|
-
- utils/enveomics/enveomics.R/man/enve.barplot.Rd
|
|
169
|
-
- utils/enveomics/enveomics.R/man/enve.cliopts.Rd
|
|
170
|
-
- utils/enveomics/enveomics.R/man/enve.col.alpha.Rd
|
|
171
|
-
- utils/enveomics/enveomics.R/man/enve.col2alpha.Rd
|
|
172
|
-
- utils/enveomics/enveomics.R/man/enve.df2dist.group.Rd
|
|
173
|
-
- utils/enveomics/enveomics.R/man/enve.df2dist.list.Rd
|
|
174
|
-
- utils/enveomics/enveomics.R/man/enve.df2dist.Rd
|
|
175
|
-
- utils/enveomics/enveomics.R/man/enve.GrowthCurve-class.Rd
|
|
176
|
-
- utils/enveomics/enveomics.R/man/enve.growthcurve.Rd
|
|
177
|
-
- utils/enveomics/enveomics.R/man/enve.prune.dist.Rd
|
|
178
|
-
- utils/enveomics/enveomics.R/man/enve.recplot.Rd
|
|
179
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2-class.Rd
|
|
180
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.__counts.Rd
|
|
181
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.__findPeak.Rd
|
|
182
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.__findPeaks.Rd
|
|
183
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.__peakHist.Rd
|
|
184
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.__whichClosestPeak.Rd
|
|
185
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.changeCutoff.Rd
|
|
186
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.corePeak.Rd
|
|
187
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.extractWindows.Rd
|
|
188
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.findPeaks.Rd
|
|
189
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.peak-class.Rd
|
|
190
|
-
- utils/enveomics/enveomics.R/man/enve.recplot2.Rd
|
|
191
|
-
- utils/enveomics/enveomics.R/man/enve.TRIBS-class.Rd
|
|
192
|
-
- utils/enveomics/enveomics.R/man/enve.TRIBS.merge.Rd
|
|
193
|
-
- utils/enveomics/enveomics.R/man/enve.tribs.Rd
|
|
194
|
-
- utils/enveomics/enveomics.R/man/enve.tribs.test.Rd
|
|
195
|
-
- utils/enveomics/enveomics.R/man/enve.TRIBStest-class.Rd
|
|
196
|
-
- utils/enveomics/enveomics.R/man/enveomics.R-package.Rd
|
|
197
|
-
- utils/enveomics/enveomics.R/man/growth.curves.Rd
|
|
198
|
-
- utils/enveomics/enveomics.R/man/phyla.counts.Rd
|
|
199
|
-
- utils/enveomics/enveomics.R/man/plot.enve.GrowthCurve.Rd
|
|
200
|
-
- utils/enveomics/enveomics.R/man/plot.enve.recplot2.Rd
|
|
201
|
-
- utils/enveomics/enveomics.R/man/plot.enve.TRIBS.Rd
|
|
202
|
-
- utils/enveomics/enveomics.R/man/plot.enve.TRIBStest.Rd
|
|
203
|
-
- utils/enveomics/enveomics.R/man/summary.enve.GrowthCurve.Rd
|
|
204
|
-
- utils/enveomics/enveomics.R/man/summary.enve.TRIBS.Rd
|
|
205
|
-
- utils/enveomics/enveomics.R/man/summary.enve.TRIBStest.Rd
|
|
206
|
-
- utils/enveomics/enveomics.R/man/z$-methods.Rd
|
|
207
|
-
- utils/enveomics/enveomics.R/NAMESPACE
|
|
208
|
-
- utils/enveomics/enveomics.R/R/autoprune.R
|
|
209
|
-
- utils/enveomics/enveomics.R/R/barplot.R
|
|
210
|
-
- utils/enveomics/enveomics.R/R/cliopts.R
|
|
211
|
-
- utils/enveomics/enveomics.R/R/df2dist.R
|
|
212
|
-
- utils/enveomics/enveomics.R/R/growthcurve.R
|
|
213
|
-
- utils/enveomics/enveomics.R/R/recplot.R
|
|
214
|
-
- utils/enveomics/enveomics.R/R/recplot2.R
|
|
215
|
-
- utils/enveomics/enveomics.R/R/tribs.R
|
|
216
|
-
- utils/enveomics/enveomics.R/R/utils.R
|
|
217
|
-
- utils/enveomics/enveomics.R/README.md
|
|
218
192
|
- utils/enveomics/Examples/aai-matrix.bash
|
|
219
193
|
- utils/enveomics/Examples/ani-matrix.bash
|
|
220
194
|
- utils/enveomics/Examples/essential-phylogeny.bash
|
|
221
195
|
- utils/enveomics/Examples/unus-genome-phylogeny.bash
|
|
222
|
-
- utils/enveomics/globals.mk
|
|
223
196
|
- utils/enveomics/LICENSE.txt
|
|
224
197
|
- utils/enveomics/Makefile
|
|
225
|
-
- utils/enveomics/Manifest/categories.json
|
|
226
|
-
- utils/enveomics/Manifest/examples.json
|
|
227
198
|
- utils/enveomics/Manifest/Tasks/aasubs.json
|
|
228
199
|
- utils/enveomics/Manifest/Tasks/blasttab.json
|
|
229
200
|
- utils/enveomics/Manifest/Tasks/distances.json
|
|
@@ -237,20 +208,21 @@ files:
|
|
|
237
208
|
- utils/enveomics/Manifest/Tasks/tables.json
|
|
238
209
|
- utils/enveomics/Manifest/Tasks/trees.json
|
|
239
210
|
- utils/enveomics/Manifest/Tasks/variants.json
|
|
211
|
+
- utils/enveomics/Manifest/categories.json
|
|
212
|
+
- utils/enveomics/Manifest/examples.json
|
|
240
213
|
- utils/enveomics/Manifest/tasks.json
|
|
241
|
-
- utils/enveomics/manifest.json
|
|
242
214
|
- utils/enveomics/Pipelines/assembly.pbs/CONFIG.mock.bash
|
|
215
|
+
- utils/enveomics/Pipelines/assembly.pbs/FastA.N50.pl
|
|
243
216
|
- utils/enveomics/Pipelines/assembly.pbs/FastA.filterN.pl
|
|
244
217
|
- utils/enveomics/Pipelines/assembly.pbs/FastA.length.pl
|
|
245
|
-
- utils/enveomics/Pipelines/assembly.pbs/FastA.N50.pl
|
|
246
|
-
- utils/enveomics/Pipelines/assembly.pbs/kSelector.R
|
|
247
|
-
- utils/enveomics/Pipelines/assembly.pbs/newbler.pbs
|
|
248
|
-
- utils/enveomics/Pipelines/assembly.pbs/newbler_preparator.pl
|
|
249
218
|
- utils/enveomics/Pipelines/assembly.pbs/README.md
|
|
250
219
|
- utils/enveomics/Pipelines/assembly.pbs/RUNME-2.bash
|
|
251
220
|
- utils/enveomics/Pipelines/assembly.pbs/RUNME-3.bash
|
|
252
221
|
- utils/enveomics/Pipelines/assembly.pbs/RUNME-4.bash
|
|
253
222
|
- utils/enveomics/Pipelines/assembly.pbs/RUNME.bash
|
|
223
|
+
- utils/enveomics/Pipelines/assembly.pbs/kSelector.R
|
|
224
|
+
- utils/enveomics/Pipelines/assembly.pbs/newbler.pbs
|
|
225
|
+
- utils/enveomics/Pipelines/assembly.pbs/newbler_preparator.pl
|
|
254
226
|
- utils/enveomics/Pipelines/assembly.pbs/soap.pbs
|
|
255
227
|
- utils/enveomics/Pipelines/assembly.pbs/stats.pbs
|
|
256
228
|
- utils/enveomics/Pipelines/assembly.pbs/velvet.pbs
|
|
@@ -262,25 +234,23 @@ files:
|
|
|
262
234
|
- utils/enveomics/Pipelines/blast.pbs/FastA.split.pl
|
|
263
235
|
- utils/enveomics/Pipelines/blast.pbs/README.md
|
|
264
236
|
- utils/enveomics/Pipelines/blast.pbs/RUNME.bash
|
|
265
|
-
- utils/enveomics/Pipelines/blast.pbs/sentinel.pbs.bash
|
|
266
237
|
- utils/enveomics/Pipelines/blast.pbs/TASK.check.bash
|
|
267
238
|
- utils/enveomics/Pipelines/blast.pbs/TASK.dry.bash
|
|
268
239
|
- utils/enveomics/Pipelines/blast.pbs/TASK.eo.bash
|
|
269
240
|
- utils/enveomics/Pipelines/blast.pbs/TASK.pause.bash
|
|
270
241
|
- utils/enveomics/Pipelines/blast.pbs/TASK.run.bash
|
|
242
|
+
- utils/enveomics/Pipelines/blast.pbs/sentinel.pbs.bash
|
|
271
243
|
- utils/enveomics/Pipelines/idba.pbs/README.md
|
|
272
|
-
- utils/enveomics/Pipelines/idba.pbs/run.pbs
|
|
273
244
|
- utils/enveomics/Pipelines/idba.pbs/RUNME.bash
|
|
245
|
+
- utils/enveomics/Pipelines/idba.pbs/run.pbs
|
|
274
246
|
- utils/enveomics/Pipelines/trim.pbs/README.md
|
|
275
|
-
- utils/enveomics/Pipelines/trim.pbs/run.pbs
|
|
276
247
|
- utils/enveomics/Pipelines/trim.pbs/RUNME.bash
|
|
248
|
+
- utils/enveomics/Pipelines/trim.pbs/run.pbs
|
|
277
249
|
- utils/enveomics/README.md
|
|
278
|
-
- utils/enveomics/Scripts/aai.rb
|
|
279
250
|
- utils/enveomics/Scripts/AAsubs.log2ratio.rb
|
|
280
251
|
- utils/enveomics/Scripts/Aln.cat.rb
|
|
281
252
|
- utils/enveomics/Scripts/Aln.convert.pl
|
|
282
253
|
- utils/enveomics/Scripts/AlphaDiversity.pl
|
|
283
|
-
- utils/enveomics/Scripts/ani.rb
|
|
284
254
|
- utils/enveomics/Scripts/BlastPairwise.AAsubs.pl
|
|
285
255
|
- utils/enveomics/Scripts/BlastTab.addlen.rb
|
|
286
256
|
- utils/enveomics/Scripts/BlastTab.advance.bash
|
|
@@ -293,8 +263,8 @@ files:
|
|
|
293
263
|
- utils/enveomics/Scripts/BlastTab.pairedHits.rb
|
|
294
264
|
- utils/enveomics/Scripts/BlastTab.recplot2.R
|
|
295
265
|
- utils/enveomics/Scripts/BlastTab.seqdepth.pl
|
|
296
|
-
- utils/enveomics/Scripts/BlastTab.seqdepth_nomedian.pl
|
|
297
266
|
- utils/enveomics/Scripts/BlastTab.seqdepth_ZIP.pl
|
|
267
|
+
- utils/enveomics/Scripts/BlastTab.seqdepth_nomedian.pl
|
|
298
268
|
- utils/enveomics/Scripts/BlastTab.subsample.pl
|
|
299
269
|
- utils/enveomics/Scripts/BlastTab.sumPerHit.pl
|
|
300
270
|
- utils/enveomics/Scripts/BlastTab.taxid2taxrank.pl
|
|
@@ -302,6 +272,7 @@ files:
|
|
|
302
272
|
- utils/enveomics/Scripts/Chao1.pl
|
|
303
273
|
- utils/enveomics/Scripts/CharTable.classify.rb
|
|
304
274
|
- utils/enveomics/Scripts/EBIseq2tax.rb
|
|
275
|
+
- utils/enveomics/Scripts/FastA.N50.pl
|
|
305
276
|
- utils/enveomics/Scripts/FastA.filter.pl
|
|
306
277
|
- utils/enveomics/Scripts/FastA.filterLen.pl
|
|
307
278
|
- utils/enveomics/Scripts/FastA.filterN.pl
|
|
@@ -309,7 +280,6 @@ files:
|
|
|
309
280
|
- utils/enveomics/Scripts/FastA.gc.pl
|
|
310
281
|
- utils/enveomics/Scripts/FastA.interpose.pl
|
|
311
282
|
- utils/enveomics/Scripts/FastA.length.pl
|
|
312
|
-
- utils/enveomics/Scripts/FastA.N50.pl
|
|
313
283
|
- utils/enveomics/Scripts/FastA.per_file.pl
|
|
314
284
|
- utils/enveomics/Scripts/FastA.qlen.pl
|
|
315
285
|
- utils/enveomics/Scripts/FastA.rename.pl
|
|
@@ -326,36 +296,21 @@ files:
|
|
|
326
296
|
- utils/enveomics/Scripts/FastQ.tag.rb
|
|
327
297
|
- utils/enveomics/Scripts/FastQ.toFastA.awk
|
|
328
298
|
- utils/enveomics/Scripts/GenBank.add_fields.rb
|
|
329
|
-
- utils/enveomics/Scripts/gi2tax.rb
|
|
330
299
|
- utils/enveomics/Scripts/HMM.essential.rb
|
|
331
300
|
- utils/enveomics/Scripts/HMMsearch.extractIds.rb
|
|
332
|
-
- utils/enveomics/Scripts/in_silico_GA_GI.pl
|
|
333
301
|
- utils/enveomics/Scripts/JPlace.distances.rb
|
|
334
302
|
- utils/enveomics/Scripts/JPlace.to_iToL.rb
|
|
335
|
-
- utils/enveomics/Scripts/lib/data/essential.hmm.gz
|
|
336
|
-
- utils/enveomics/Scripts/lib/enveomics_rb/enveomics.rb
|
|
337
|
-
- utils/enveomics/Scripts/lib/enveomics_rb/jplace.rb
|
|
338
|
-
- utils/enveomics/Scripts/lib/enveomics_rb/og.rb
|
|
339
|
-
- utils/enveomics/Scripts/lib/enveomics_rb/remote_data.rb
|
|
340
|
-
- utils/enveomics/Scripts/lib/enveomics_rb/seq_range.rb
|
|
341
|
-
- utils/enveomics/Scripts/lib/enveomics_rb/stat.rb
|
|
342
|
-
- utils/enveomics/Scripts/lib/enveomics_rb/vcf.rb
|
|
343
303
|
- utils/enveomics/Scripts/M5nr.getSequences.rb
|
|
344
304
|
- utils/enveomics/Scripts/MeTaxa.distribution.pl
|
|
345
305
|
- utils/enveomics/Scripts/MyTaxa.fragsByTax.pl
|
|
346
306
|
- utils/enveomics/Scripts/MyTaxa.seq-taxrank.rb
|
|
347
307
|
- utils/enveomics/Scripts/NCBIacc2tax.rb
|
|
348
308
|
- utils/enveomics/Scripts/Newick.autoprune.R
|
|
349
|
-
- utils/enveomics/Scripts/ogs.annotate.rb
|
|
350
|
-
- utils/enveomics/Scripts/ogs.core-pan.rb
|
|
351
|
-
- utils/enveomics/Scripts/ogs.extract.rb
|
|
352
|
-
- utils/enveomics/Scripts/ogs.mcl.rb
|
|
353
|
-
- utils/enveomics/Scripts/ogs.rb
|
|
354
|
-
- utils/enveomics/Scripts/ogs.stats.rb
|
|
355
309
|
- utils/enveomics/Scripts/RAxML-EPA.to_iToL.pl
|
|
356
|
-
- utils/enveomics/Scripts/rbm.rb
|
|
357
310
|
- utils/enveomics/Scripts/RefSeq.download.bash
|
|
358
311
|
- utils/enveomics/Scripts/SRA.download.bash
|
|
312
|
+
- utils/enveomics/Scripts/TRIBS.plot-test.R
|
|
313
|
+
- utils/enveomics/Scripts/TRIBS.test.R
|
|
359
314
|
- utils/enveomics/Scripts/Table.barplot.R
|
|
360
315
|
- utils/enveomics/Scripts/Table.df2dist.R
|
|
361
316
|
- utils/enveomics/Scripts/Table.filter.pl
|
|
@@ -364,10 +319,37 @@ files:
|
|
|
364
319
|
- utils/enveomics/Scripts/Table.round.rb
|
|
365
320
|
- utils/enveomics/Scripts/Table.split.pl
|
|
366
321
|
- utils/enveomics/Scripts/Taxonomy.silva2ncbi.rb
|
|
367
|
-
- utils/enveomics/Scripts/TRIBS.plot-test.R
|
|
368
|
-
- utils/enveomics/Scripts/TRIBS.test.R
|
|
369
322
|
- utils/enveomics/Scripts/VCF.KaKs.rb
|
|
370
323
|
- utils/enveomics/Scripts/VCF.SNPs.rb
|
|
324
|
+
- utils/enveomics/Scripts/aai.rb
|
|
325
|
+
- utils/enveomics/Scripts/ani.rb
|
|
326
|
+
- utils/enveomics/Scripts/gi2tax.rb
|
|
327
|
+
- utils/enveomics/Scripts/in_silico_GA_GI.pl
|
|
328
|
+
- utils/enveomics/Scripts/lib/data/essential.hmm.gz
|
|
329
|
+
- utils/enveomics/Scripts/lib/enveomics.R
|
|
330
|
+
- utils/enveomics/Scripts/lib/enveomics_rb/enveomics.rb
|
|
331
|
+
- utils/enveomics/Scripts/lib/enveomics_rb/jplace.rb
|
|
332
|
+
- utils/enveomics/Scripts/lib/enveomics_rb/og.rb
|
|
333
|
+
- utils/enveomics/Scripts/lib/enveomics_rb/remote_data.rb
|
|
334
|
+
- utils/enveomics/Scripts/lib/enveomics_rb/seq_range.rb
|
|
335
|
+
- utils/enveomics/Scripts/lib/enveomics_rb/stat.rb
|
|
336
|
+
- utils/enveomics/Scripts/lib/enveomics_rb/vcf.rb
|
|
337
|
+
- utils/enveomics/Scripts/ogs.annotate.rb
|
|
338
|
+
- utils/enveomics/Scripts/ogs.core-pan.rb
|
|
339
|
+
- utils/enveomics/Scripts/ogs.extract.rb
|
|
340
|
+
- utils/enveomics/Scripts/ogs.mcl.rb
|
|
341
|
+
- utils/enveomics/Scripts/ogs.rb
|
|
342
|
+
- utils/enveomics/Scripts/ogs.stats.rb
|
|
343
|
+
- utils/enveomics/Scripts/rbm.rb
|
|
344
|
+
- utils/enveomics/Tests/Makefile
|
|
345
|
+
- utils/enveomics/Tests/Mgen_M2288.faa
|
|
346
|
+
- utils/enveomics/Tests/Mgen_M2288.fna
|
|
347
|
+
- utils/enveomics/Tests/Mgen_M2321.fna
|
|
348
|
+
- utils/enveomics/Tests/Nequ_Kin4M.faa
|
|
349
|
+
- utils/enveomics/Tests/Xanthomonas_oryzae-PilA.tribs.Rdata
|
|
350
|
+
- utils/enveomics/Tests/Xanthomonas_oryzae-PilA.txt
|
|
351
|
+
- utils/enveomics/Tests/Xanthomonas_oryzae.aai-mat.tsv
|
|
352
|
+
- utils/enveomics/Tests/Xanthomonas_oryzae.aai.tsv
|
|
371
353
|
- utils/enveomics/Tests/a_mg.cds-go.blast.tsv
|
|
372
354
|
- utils/enveomics/Tests/a_mg.reads-cds.blast.tsv
|
|
373
355
|
- utils/enveomics/Tests/a_mg.reads-cds.counts.tsv
|
|
@@ -380,11 +362,6 @@ files:
|
|
|
380
362
|
- utils/enveomics/Tests/hiv_mix-hiv1.blast.tsv
|
|
381
363
|
- utils/enveomics/Tests/hiv_mix-hiv1.blast.tsv.lim
|
|
382
364
|
- utils/enveomics/Tests/hiv_mix-hiv1.blast.tsv.rec
|
|
383
|
-
- utils/enveomics/Tests/Makefile
|
|
384
|
-
- utils/enveomics/Tests/Mgen_M2288.faa
|
|
385
|
-
- utils/enveomics/Tests/Mgen_M2288.fna
|
|
386
|
-
- utils/enveomics/Tests/Mgen_M2321.fna
|
|
387
|
-
- utils/enveomics/Tests/Nequ_Kin4M.faa
|
|
388
365
|
- utils/enveomics/Tests/phyla_counts.tsv
|
|
389
366
|
- utils/enveomics/Tests/primate_lentivirus.ogs
|
|
390
367
|
- utils/enveomics/Tests/primate_lentivirus.rbm/hiv1-hiv1.rbm
|
|
@@ -393,10 +370,65 @@ files:
|
|
|
393
370
|
- utils/enveomics/Tests/primate_lentivirus.rbm/hiv2-hiv2.rbm
|
|
394
371
|
- utils/enveomics/Tests/primate_lentivirus.rbm/hiv2-siv.rbm
|
|
395
372
|
- utils/enveomics/Tests/primate_lentivirus.rbm/siv-siv.rbm
|
|
396
|
-
- utils/enveomics/
|
|
397
|
-
- utils/enveomics/
|
|
398
|
-
- utils/enveomics/
|
|
399
|
-
- utils/enveomics/
|
|
373
|
+
- utils/enveomics/build_enveomics_r.bash
|
|
374
|
+
- utils/enveomics/enveomics.R/DESCRIPTION
|
|
375
|
+
- utils/enveomics/enveomics.R/NAMESPACE
|
|
376
|
+
- utils/enveomics/enveomics.R/R/autoprune.R
|
|
377
|
+
- utils/enveomics/enveomics.R/R/barplot.R
|
|
378
|
+
- utils/enveomics/enveomics.R/R/cliopts.R
|
|
379
|
+
- utils/enveomics/enveomics.R/R/df2dist.R
|
|
380
|
+
- utils/enveomics/enveomics.R/R/growthcurve.R
|
|
381
|
+
- utils/enveomics/enveomics.R/R/recplot.R
|
|
382
|
+
- utils/enveomics/enveomics.R/R/recplot2.R
|
|
383
|
+
- utils/enveomics/enveomics.R/R/tribs.R
|
|
384
|
+
- utils/enveomics/enveomics.R/R/utils.R
|
|
385
|
+
- utils/enveomics/enveomics.R/README.md
|
|
386
|
+
- utils/enveomics/enveomics.R/data/growth.curves.rda
|
|
387
|
+
- utils/enveomics/enveomics.R/data/phyla.counts.rda
|
|
388
|
+
- utils/enveomics/enveomics.R/man/enve.GrowthCurve-class.Rd
|
|
389
|
+
- utils/enveomics/enveomics.R/man/enve.TRIBS-class.Rd
|
|
390
|
+
- utils/enveomics/enveomics.R/man/enve.TRIBS.merge.Rd
|
|
391
|
+
- utils/enveomics/enveomics.R/man/enve.TRIBStest-class.Rd
|
|
392
|
+
- utils/enveomics/enveomics.R/man/enve.__prune.iter.Rd
|
|
393
|
+
- utils/enveomics/enveomics.R/man/enve.__prune.reduce.Rd
|
|
394
|
+
- utils/enveomics/enveomics.R/man/enve.__tribs.Rd
|
|
395
|
+
- utils/enveomics/enveomics.R/man/enve.barplot.Rd
|
|
396
|
+
- utils/enveomics/enveomics.R/man/enve.cliopts.Rd
|
|
397
|
+
- utils/enveomics/enveomics.R/man/enve.col.alpha.Rd
|
|
398
|
+
- utils/enveomics/enveomics.R/man/enve.col2alpha.Rd
|
|
399
|
+
- utils/enveomics/enveomics.R/man/enve.df2dist.Rd
|
|
400
|
+
- utils/enveomics/enveomics.R/man/enve.df2dist.group.Rd
|
|
401
|
+
- utils/enveomics/enveomics.R/man/enve.df2dist.list.Rd
|
|
402
|
+
- utils/enveomics/enveomics.R/man/enve.growthcurve.Rd
|
|
403
|
+
- utils/enveomics/enveomics.R/man/enve.prune.dist.Rd
|
|
404
|
+
- utils/enveomics/enveomics.R/man/enve.recplot.Rd
|
|
405
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2-class.Rd
|
|
406
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.Rd
|
|
407
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.__counts.Rd
|
|
408
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.__findPeak.Rd
|
|
409
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.__findPeaks.Rd
|
|
410
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.__peakHist.Rd
|
|
411
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.__whichClosestPeak.Rd
|
|
412
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.changeCutoff.Rd
|
|
413
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.corePeak.Rd
|
|
414
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.extractWindows.Rd
|
|
415
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.findPeaks.Rd
|
|
416
|
+
- utils/enveomics/enveomics.R/man/enve.recplot2.peak-class.Rd
|
|
417
|
+
- utils/enveomics/enveomics.R/man/enve.tribs.Rd
|
|
418
|
+
- utils/enveomics/enveomics.R/man/enve.tribs.test.Rd
|
|
419
|
+
- utils/enveomics/enveomics.R/man/enveomics.R-package.Rd
|
|
420
|
+
- utils/enveomics/enveomics.R/man/growth.curves.Rd
|
|
421
|
+
- utils/enveomics/enveomics.R/man/phyla.counts.Rd
|
|
422
|
+
- utils/enveomics/enveomics.R/man/plot.enve.GrowthCurve.Rd
|
|
423
|
+
- utils/enveomics/enveomics.R/man/plot.enve.TRIBS.Rd
|
|
424
|
+
- utils/enveomics/enveomics.R/man/plot.enve.TRIBStest.Rd
|
|
425
|
+
- utils/enveomics/enveomics.R/man/plot.enve.recplot2.Rd
|
|
426
|
+
- utils/enveomics/enveomics.R/man/summary.enve.GrowthCurve.Rd
|
|
427
|
+
- utils/enveomics/enveomics.R/man/summary.enve.TRIBS.Rd
|
|
428
|
+
- utils/enveomics/enveomics.R/man/summary.enve.TRIBStest.Rd
|
|
429
|
+
- utils/enveomics/enveomics.R/man/z$-methods.Rd
|
|
430
|
+
- utils/enveomics/globals.mk
|
|
431
|
+
- utils/enveomics/manifest.json
|
|
400
432
|
- utils/index_metadata.rb
|
|
401
433
|
- utils/mytaxa_scan.R
|
|
402
434
|
- utils/mytaxa_scan.rb
|
|
@@ -406,31 +438,6 @@ files:
|
|
|
406
438
|
- utils/subclades-compile.rb
|
|
407
439
|
- utils/subclades-nj.R
|
|
408
440
|
- utils/subclades.R
|
|
409
|
-
- bin/miga
|
|
410
|
-
- actions/about.rb
|
|
411
|
-
- actions/add.rb
|
|
412
|
-
- actions/add_result.rb
|
|
413
|
-
- actions/daemon.rb
|
|
414
|
-
- actions/date.rb
|
|
415
|
-
- actions/files.rb
|
|
416
|
-
- actions/find.rb
|
|
417
|
-
- actions/get.rb
|
|
418
|
-
- actions/init.rb
|
|
419
|
-
- actions/ln.rb
|
|
420
|
-
- actions/ls.rb
|
|
421
|
-
- actions/new.rb
|
|
422
|
-
- actions/plugins.rb
|
|
423
|
-
- actions/rm.rb
|
|
424
|
-
- actions/run.rb
|
|
425
|
-
- actions/stats.rb
|
|
426
|
-
- actions/tax_dist.rb
|
|
427
|
-
- actions/tax_index.rb
|
|
428
|
-
- actions/tax_set.rb
|
|
429
|
-
- actions/tax_test.rb
|
|
430
|
-
- Gemfile
|
|
431
|
-
- Rakefile
|
|
432
|
-
- README.md
|
|
433
|
-
- LICENSE
|
|
434
441
|
homepage: http://enve-omics.ce.gatech.edu/miga
|
|
435
442
|
licenses:
|
|
436
443
|
- Artistic-2.0
|
|
@@ -439,25 +446,25 @@ post_install_message:
|
|
|
439
446
|
rdoc_options:
|
|
440
447
|
- lib
|
|
441
448
|
- README.md
|
|
442
|
-
- --main
|
|
449
|
+
- "--main"
|
|
443
450
|
- README.md
|
|
444
|
-
- --title
|
|
451
|
+
- "--title"
|
|
445
452
|
- MiGA
|
|
446
453
|
require_paths:
|
|
447
454
|
- lib
|
|
448
455
|
required_ruby_version: !ruby/object:Gem::Requirement
|
|
449
456
|
requirements:
|
|
450
|
-
- -
|
|
457
|
+
- - ">="
|
|
451
458
|
- !ruby/object:Gem::Version
|
|
452
459
|
version: '1.9'
|
|
453
460
|
required_rubygems_version: !ruby/object:Gem::Requirement
|
|
454
461
|
requirements:
|
|
455
|
-
- -
|
|
462
|
+
- - ">="
|
|
456
463
|
- !ruby/object:Gem::Version
|
|
457
464
|
version: '0'
|
|
458
465
|
requirements: []
|
|
459
466
|
rubyforge_project:
|
|
460
|
-
rubygems_version: 2.
|
|
467
|
+
rubygems_version: 2.6.11
|
|
461
468
|
signing_key:
|
|
462
469
|
specification_version: 4
|
|
463
470
|
summary: MiGA
|
|
@@ -1,56 +0,0 @@
|
|
|
1
|
-
#!/usr/bin/env perl
|
|
2
|
-
#
|
|
3
|
-
# @author: Luis M. Rodriguez-R <lmrodriguezr at gmail dot com>
|
|
4
|
-
# @update: Oct 07 2015
|
|
5
|
-
# @license: artistic license 2.0
|
|
6
|
-
#
|
|
7
|
-
use strict;
|
|
8
|
-
use warnings;
|
|
9
|
-
use List::Util qw/sum min max/;
|
|
10
|
-
|
|
11
|
-
my ($seqs, $minlen, $n__) = @ARGV;
|
|
12
|
-
$seqs or die "
|
|
13
|
-
Description:
|
|
14
|
-
Calculates the N50 value of a set of sequences. Alternatively, it
|
|
15
|
-
can calculate other N** values. It also calculates the total number
|
|
16
|
-
of sequences and the total added length.
|
|
17
|
-
|
|
18
|
-
Usage:
|
|
19
|
-
$0 seqs.fa[ minlen[ **]]
|
|
20
|
-
|
|
21
|
-
seqs.fa A FastA file containing the sequences.
|
|
22
|
-
minlen (optional) The minimum length to take into consideration.
|
|
23
|
-
By default: 0.
|
|
24
|
-
** Value N** to calculate. By default: 50 (N50).
|
|
25
|
-
";
|
|
26
|
-
$minlen ||= 0;
|
|
27
|
-
$n__ ||= 50;
|
|
28
|
-
|
|
29
|
-
my @len = ();
|
|
30
|
-
open SEQ, "<", $seqs or die "Cannot open file: $seqs: $!\n";
|
|
31
|
-
while(<SEQ>){
|
|
32
|
-
if(/^>/){
|
|
33
|
-
push @len, 0;
|
|
34
|
-
}else{
|
|
35
|
-
next if /^;/;
|
|
36
|
-
chomp;
|
|
37
|
-
s/\W//g;
|
|
38
|
-
$len[-1]+=length $_;
|
|
39
|
-
}
|
|
40
|
-
}
|
|
41
|
-
close SEQ;
|
|
42
|
-
@len = sort { $a <=> $b } map { $_>=$minlen?$_:() } @len;
|
|
43
|
-
my $tot = (sum(@len) || 0);
|
|
44
|
-
|
|
45
|
-
my $thr = $n__*$tot/100;
|
|
46
|
-
my $pos = 0;
|
|
47
|
-
for(@len){
|
|
48
|
-
$pos+= $_;
|
|
49
|
-
if($pos>=$thr){
|
|
50
|
-
print "N$n__: $_\n";
|
|
51
|
-
last;
|
|
52
|
-
}
|
|
53
|
-
}
|
|
54
|
-
print "Sequences: ".scalar(@len)."\n";
|
|
55
|
-
print "Total length: $tot\n";
|
|
56
|
-
|
|
@@ -1,60 +0,0 @@
|
|
|
1
|
-
#!/usr/bin/env perl
|
|
2
|
-
#
|
|
3
|
-
# @author Luis M. Rodriguez-R
|
|
4
|
-
# @update Oct-07-2015
|
|
5
|
-
# @license artistic license 2.0
|
|
6
|
-
#
|
|
7
|
-
|
|
8
|
-
use warnings;
|
|
9
|
-
use strict;
|
|
10
|
-
|
|
11
|
-
my($file, $content, $stretch) = @ARGV;
|
|
12
|
-
$file or die <<HELP
|
|
13
|
-
|
|
14
|
-
Description:
|
|
15
|
-
Filter sequences by N-content and presence of long homopolymers.
|
|
16
|
-
Usage:
|
|
17
|
-
$0 sequences.fa [content [stretch]] > filtered.fa
|
|
18
|
-
Where:
|
|
19
|
-
sequences.fa Input file in FastA format
|
|
20
|
-
content A number between 0 and 1 indicating the maximum proportion of Ns
|
|
21
|
-
(1 to turn off, 0.5 by default)
|
|
22
|
-
stretch A number indicating the maximum number of consecutive identical
|
|
23
|
-
nucleotides allowed (0 to turn off, 100 by default)
|
|
24
|
-
filtered.fa Filtered set of sequences.
|
|
25
|
-
|
|
26
|
-
HELP
|
|
27
|
-
;
|
|
28
|
-
($content ||= 0.5)+=0;
|
|
29
|
-
($stretch ||= 100)+=0;
|
|
30
|
-
|
|
31
|
-
my $good = 0;
|
|
32
|
-
my $N = 0;
|
|
33
|
-
|
|
34
|
-
FASTA: {
|
|
35
|
-
local $/ = "\n>";
|
|
36
|
-
open FILE, "<", $file or die "I can not open the file: $file: $!\n";
|
|
37
|
-
SEQ: while(<FILE>){
|
|
38
|
-
$N++;
|
|
39
|
-
s/^;.*//gm;
|
|
40
|
-
s/>//g;
|
|
41
|
-
my($n,$s) = split /\n/, $_, 2;
|
|
42
|
-
(my $clean = $s) =~ s/[^ACTGN]//g;
|
|
43
|
-
if($content < 1){
|
|
44
|
-
(my $Ns = $clean) =~ s/[^N]//g;
|
|
45
|
-
next SEQ if length($Ns)>length($clean)*$content;
|
|
46
|
-
}
|
|
47
|
-
if($stretch > 0){
|
|
48
|
-
for my $nuc (qw(A C T G N)){
|
|
49
|
-
next SEQ if $clean =~ m/[$nuc]{$stretch}/;
|
|
50
|
-
}
|
|
51
|
-
}
|
|
52
|
-
print ">$n\n$s\n";
|
|
53
|
-
$good++;
|
|
54
|
-
}
|
|
55
|
-
close FILE;
|
|
56
|
-
print STDERR "Total sequences: $N\nAfter filtering: $good\n";
|
|
57
|
-
}
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
|
|
@@ -1,38 +0,0 @@
|
|
|
1
|
-
#!/usr/bin/env perl
|
|
2
|
-
#
|
|
3
|
-
# @author Luis M Rodriguez-R
|
|
4
|
-
# @update Oct-07-2015
|
|
5
|
-
# @license artistic license 2.0
|
|
6
|
-
#
|
|
7
|
-
|
|
8
|
-
use warnings;
|
|
9
|
-
use strict;
|
|
10
|
-
|
|
11
|
-
$#ARGV>=0 or die "
|
|
12
|
-
Usage:
|
|
13
|
-
$0 seqs.fa... > length.txt
|
|
14
|
-
|
|
15
|
-
seqs.fa One or more FastA files.
|
|
16
|
-
length.txt A table with the lengths of the sequences.
|
|
17
|
-
|
|
18
|
-
";
|
|
19
|
-
|
|
20
|
-
for my $fa (@ARGV){
|
|
21
|
-
open FA, "<", $fa or die "Cannot open file: $fa: $!\n";
|
|
22
|
-
my $def = '';
|
|
23
|
-
my $len = 0;
|
|
24
|
-
while(<FA>){
|
|
25
|
-
next if /^;/;
|
|
26
|
-
if(m/^>(\S+)\s?/){
|
|
27
|
-
print "$def\t$len\n" if $def;
|
|
28
|
-
$def = $1;
|
|
29
|
-
$len = 0;
|
|
30
|
-
}else{
|
|
31
|
-
s/[^A-Za-z]//g;
|
|
32
|
-
$len+= length $_;
|
|
33
|
-
}
|
|
34
|
-
}
|
|
35
|
-
print "$def\t$len\n" if $def;
|
|
36
|
-
close FA;
|
|
37
|
-
}
|
|
38
|
-
|
|
@@ -1,55 +0,0 @@
|
|
|
1
|
-
#!/usr/bin/env perl
|
|
2
|
-
#
|
|
3
|
-
# @author Luis M. Rodriguez-R <lmrodriguezr at gmail dot com>
|
|
4
|
-
# @update Oct-13-2015
|
|
5
|
-
# @license artistic license 2.0
|
|
6
|
-
#
|
|
7
|
-
|
|
8
|
-
use warnings;
|
|
9
|
-
use strict;
|
|
10
|
-
use Symbol;
|
|
11
|
-
|
|
12
|
-
my ($file, $base, $outN) = @ARGV;
|
|
13
|
-
|
|
14
|
-
$outN ||= 12;
|
|
15
|
-
($file and $base) or die "
|
|
16
|
-
Usage
|
|
17
|
-
$0 in_file.fa out_base[ no_files]
|
|
18
|
-
|
|
19
|
-
in_file.fa Input file in FastA format.
|
|
20
|
-
out_base Prefix for the name of the output files. It will
|
|
21
|
-
be appended with .<i>.fa, where <i> is a consecutive
|
|
22
|
-
number starting in 1.
|
|
23
|
-
no_files Number of files to generate. By default: 12.
|
|
24
|
-
|
|
25
|
-
";
|
|
26
|
-
|
|
27
|
-
|
|
28
|
-
my @outSym = ();
|
|
29
|
-
for my $i (1 .. $outN){
|
|
30
|
-
$outSym[$i-1] = gensym;
|
|
31
|
-
open $outSym[$i-1], ">", "$base.$i.fa" or
|
|
32
|
-
die "I can not create the file: $base.$i.fa: $!\n";
|
|
33
|
-
}
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
my($i, $seq) = (-1, '');
|
|
37
|
-
open FILE, "<", $file or die "I can not read the file: $file: $!\n";
|
|
38
|
-
while(my $ln=<FILE>){
|
|
39
|
-
next if $ln=~/^;/;
|
|
40
|
-
if($ln =~ m/^>/){
|
|
41
|
-
print { $outSym[$i % $outN] } $seq if $seq;
|
|
42
|
-
$i++;
|
|
43
|
-
$seq = '';
|
|
44
|
-
}
|
|
45
|
-
$seq.=$ln;
|
|
46
|
-
}
|
|
47
|
-
print { $outSym[$i % $outN] } $seq if $seq;
|
|
48
|
-
close FILE;
|
|
49
|
-
|
|
50
|
-
for(my $j=0; $j<$outN; $j++){
|
|
51
|
-
close $outSym[$j];
|
|
52
|
-
}
|
|
53
|
-
|
|
54
|
-
print STDERR "Sequences: ".($i+1)."\nFiles: $outN\n";
|
|
55
|
-
|