BioDSL 1.0.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +10 -0
- data/BioDSL.gemspec +64 -0
- data/LICENSE +339 -0
- data/README.md +205 -0
- data/Rakefile +94 -0
- data/examples/fastq_to_fasta.rb +8 -0
- data/lib/BioDSL/cary.rb +242 -0
- data/lib/BioDSL/command.rb +133 -0
- data/lib/BioDSL/commands/add_key.rb +110 -0
- data/lib/BioDSL/commands/align_seq_mothur.rb +194 -0
- data/lib/BioDSL/commands/analyze_residue_distribution.rb +222 -0
- data/lib/BioDSL/commands/assemble_pairs.rb +336 -0
- data/lib/BioDSL/commands/assemble_seq_idba.rb +230 -0
- data/lib/BioDSL/commands/assemble_seq_ray.rb +345 -0
- data/lib/BioDSL/commands/assemble_seq_spades.rb +252 -0
- data/lib/BioDSL/commands/classify_seq.rb +217 -0
- data/lib/BioDSL/commands/classify_seq_mothur.rb +226 -0
- data/lib/BioDSL/commands/clip_primer.rb +318 -0
- data/lib/BioDSL/commands/cluster_otus.rb +181 -0
- data/lib/BioDSL/commands/collapse_otus.rb +170 -0
- data/lib/BioDSL/commands/collect_otus.rb +150 -0
- data/lib/BioDSL/commands/complement_seq.rb +117 -0
- data/lib/BioDSL/commands/count.rb +135 -0
- data/lib/BioDSL/commands/count_values.rb +149 -0
- data/lib/BioDSL/commands/degap_seq.rb +253 -0
- data/lib/BioDSL/commands/dereplicate_seq.rb +168 -0
- data/lib/BioDSL/commands/dump.rb +157 -0
- data/lib/BioDSL/commands/filter_rrna.rb +239 -0
- data/lib/BioDSL/commands/genecall.rb +237 -0
- data/lib/BioDSL/commands/grab.rb +535 -0
- data/lib/BioDSL/commands/index_taxonomy.rb +226 -0
- data/lib/BioDSL/commands/mask_seq.rb +175 -0
- data/lib/BioDSL/commands/mean_scores.rb +168 -0
- data/lib/BioDSL/commands/merge_pair_seq.rb +175 -0
- data/lib/BioDSL/commands/merge_table.rb +225 -0
- data/lib/BioDSL/commands/merge_values.rb +113 -0
- data/lib/BioDSL/commands/plot_heatmap.rb +233 -0
- data/lib/BioDSL/commands/plot_histogram.rb +306 -0
- data/lib/BioDSL/commands/plot_matches.rb +282 -0
- data/lib/BioDSL/commands/plot_residue_distribution.rb +278 -0
- data/lib/BioDSL/commands/plot_scores.rb +285 -0
- data/lib/BioDSL/commands/random.rb +153 -0
- data/lib/BioDSL/commands/read_fasta.rb +222 -0
- data/lib/BioDSL/commands/read_fastq.rb +414 -0
- data/lib/BioDSL/commands/read_table.rb +329 -0
- data/lib/BioDSL/commands/reverse_seq.rb +113 -0
- data/lib/BioDSL/commands/slice_align.rb +400 -0
- data/lib/BioDSL/commands/slice_seq.rb +151 -0
- data/lib/BioDSL/commands/sort.rb +223 -0
- data/lib/BioDSL/commands/split_pair_seq.rb +220 -0
- data/lib/BioDSL/commands/split_values.rb +165 -0
- data/lib/BioDSL/commands/trim_primer.rb +314 -0
- data/lib/BioDSL/commands/trim_seq.rb +192 -0
- data/lib/BioDSL/commands/uchime_ref.rb +170 -0
- data/lib/BioDSL/commands/uclust.rb +286 -0
- data/lib/BioDSL/commands/unique_values.rb +145 -0
- data/lib/BioDSL/commands/usearch_global.rb +171 -0
- data/lib/BioDSL/commands/usearch_local.rb +171 -0
- data/lib/BioDSL/commands/write_fasta.rb +207 -0
- data/lib/BioDSL/commands/write_fastq.rb +191 -0
- data/lib/BioDSL/commands/write_table.rb +419 -0
- data/lib/BioDSL/commands/write_tree.rb +167 -0
- data/lib/BioDSL/commands.rb +31 -0
- data/lib/BioDSL/config.rb +55 -0
- data/lib/BioDSL/csv.rb +307 -0
- data/lib/BioDSL/debug.rb +42 -0
- data/lib/BioDSL/fasta.rb +133 -0
- data/lib/BioDSL/fastq.rb +77 -0
- data/lib/BioDSL/filesys.rb +137 -0
- data/lib/BioDSL/fork.rb +145 -0
- data/lib/BioDSL/hamming.rb +128 -0
- data/lib/BioDSL/helpers/aux_helper.rb +44 -0
- data/lib/BioDSL/helpers/email_helper.rb +66 -0
- data/lib/BioDSL/helpers/history_helper.rb +40 -0
- data/lib/BioDSL/helpers/log_helper.rb +55 -0
- data/lib/BioDSL/helpers/options_helper.rb +405 -0
- data/lib/BioDSL/helpers/status_helper.rb +132 -0
- data/lib/BioDSL/helpers.rb +35 -0
- data/lib/BioDSL/html_report.rb +200 -0
- data/lib/BioDSL/math.rb +55 -0
- data/lib/BioDSL/mummer.rb +216 -0
- data/lib/BioDSL/pipeline.rb +354 -0
- data/lib/BioDSL/seq/ambiguity.rb +66 -0
- data/lib/BioDSL/seq/assemble.rb +240 -0
- data/lib/BioDSL/seq/backtrack.rb +252 -0
- data/lib/BioDSL/seq/digest.rb +99 -0
- data/lib/BioDSL/seq/dynamic.rb +263 -0
- data/lib/BioDSL/seq/homopolymer.rb +59 -0
- data/lib/BioDSL/seq/kmer.rb +293 -0
- data/lib/BioDSL/seq/levenshtein.rb +113 -0
- data/lib/BioDSL/seq/translate.rb +109 -0
- data/lib/BioDSL/seq/trim.rb +188 -0
- data/lib/BioDSL/seq.rb +742 -0
- data/lib/BioDSL/serializer.rb +98 -0
- data/lib/BioDSL/stream.rb +113 -0
- data/lib/BioDSL/taxonomy.rb +691 -0
- data/lib/BioDSL/test.rb +42 -0
- data/lib/BioDSL/tmp_dir.rb +68 -0
- data/lib/BioDSL/usearch.rb +301 -0
- data/lib/BioDSL/verbose.rb +42 -0
- data/lib/BioDSL/version.rb +31 -0
- data/lib/BioDSL.rb +81 -0
- data/test/BioDSL/commands/test_add_key.rb +105 -0
- data/test/BioDSL/commands/test_align_seq_mothur.rb +99 -0
- data/test/BioDSL/commands/test_analyze_residue_distribution.rb +134 -0
- data/test/BioDSL/commands/test_assemble_pairs.rb +459 -0
- data/test/BioDSL/commands/test_assemble_seq_idba.rb +50 -0
- data/test/BioDSL/commands/test_assemble_seq_ray.rb +51 -0
- data/test/BioDSL/commands/test_assemble_seq_spades.rb +50 -0
- data/test/BioDSL/commands/test_classify_seq.rb +50 -0
- data/test/BioDSL/commands/test_classify_seq_mothur.rb +59 -0
- data/test/BioDSL/commands/test_clip_primer.rb +377 -0
- data/test/BioDSL/commands/test_cluster_otus.rb +128 -0
- data/test/BioDSL/commands/test_collapse_otus.rb +81 -0
- data/test/BioDSL/commands/test_collect_otus.rb +82 -0
- data/test/BioDSL/commands/test_complement_seq.rb +78 -0
- data/test/BioDSL/commands/test_count.rb +103 -0
- data/test/BioDSL/commands/test_count_values.rb +85 -0
- data/test/BioDSL/commands/test_degap_seq.rb +96 -0
- data/test/BioDSL/commands/test_dereplicate_seq.rb +92 -0
- data/test/BioDSL/commands/test_dump.rb +109 -0
- data/test/BioDSL/commands/test_filter_rrna.rb +128 -0
- data/test/BioDSL/commands/test_genecall.rb +50 -0
- data/test/BioDSL/commands/test_grab.rb +398 -0
- data/test/BioDSL/commands/test_index_taxonomy.rb +62 -0
- data/test/BioDSL/commands/test_mask_seq.rb +98 -0
- data/test/BioDSL/commands/test_mean_scores.rb +111 -0
- data/test/BioDSL/commands/test_merge_pair_seq.rb +115 -0
- data/test/BioDSL/commands/test_merge_table.rb +131 -0
- data/test/BioDSL/commands/test_merge_values.rb +83 -0
- data/test/BioDSL/commands/test_plot_heatmap.rb +185 -0
- data/test/BioDSL/commands/test_plot_histogram.rb +194 -0
- data/test/BioDSL/commands/test_plot_matches.rb +157 -0
- data/test/BioDSL/commands/test_plot_residue_distribution.rb +309 -0
- data/test/BioDSL/commands/test_plot_scores.rb +308 -0
- data/test/BioDSL/commands/test_random.rb +88 -0
- data/test/BioDSL/commands/test_read_fasta.rb +229 -0
- data/test/BioDSL/commands/test_read_fastq.rb +552 -0
- data/test/BioDSL/commands/test_read_table.rb +327 -0
- data/test/BioDSL/commands/test_reverse_seq.rb +79 -0
- data/test/BioDSL/commands/test_slice_align.rb +218 -0
- data/test/BioDSL/commands/test_slice_seq.rb +131 -0
- data/test/BioDSL/commands/test_sort.rb +128 -0
- data/test/BioDSL/commands/test_split_pair_seq.rb +164 -0
- data/test/BioDSL/commands/test_split_values.rb +95 -0
- data/test/BioDSL/commands/test_trim_primer.rb +329 -0
- data/test/BioDSL/commands/test_trim_seq.rb +150 -0
- data/test/BioDSL/commands/test_uchime_ref.rb +113 -0
- data/test/BioDSL/commands/test_uclust.rb +139 -0
- data/test/BioDSL/commands/test_unique_values.rb +98 -0
- data/test/BioDSL/commands/test_usearch_global.rb +123 -0
- data/test/BioDSL/commands/test_usearch_local.rb +125 -0
- data/test/BioDSL/commands/test_write_fasta.rb +159 -0
- data/test/BioDSL/commands/test_write_fastq.rb +166 -0
- data/test/BioDSL/commands/test_write_table.rb +411 -0
- data/test/BioDSL/commands/test_write_tree.rb +122 -0
- data/test/BioDSL/helpers/test_options_helper.rb +272 -0
- data/test/BioDSL/seq/test_assemble.rb +98 -0
- data/test/BioDSL/seq/test_backtrack.rb +176 -0
- data/test/BioDSL/seq/test_digest.rb +71 -0
- data/test/BioDSL/seq/test_dynamic.rb +133 -0
- data/test/BioDSL/seq/test_homopolymer.rb +58 -0
- data/test/BioDSL/seq/test_kmer.rb +134 -0
- data/test/BioDSL/seq/test_translate.rb +75 -0
- data/test/BioDSL/seq/test_trim.rb +101 -0
- data/test/BioDSL/test_cary.rb +176 -0
- data/test/BioDSL/test_command.rb +45 -0
- data/test/BioDSL/test_csv.rb +514 -0
- data/test/BioDSL/test_debug.rb +42 -0
- data/test/BioDSL/test_fasta.rb +154 -0
- data/test/BioDSL/test_fastq.rb +46 -0
- data/test/BioDSL/test_filesys.rb +145 -0
- data/test/BioDSL/test_fork.rb +85 -0
- data/test/BioDSL/test_math.rb +41 -0
- data/test/BioDSL/test_mummer.rb +79 -0
- data/test/BioDSL/test_pipeline.rb +187 -0
- data/test/BioDSL/test_seq.rb +790 -0
- data/test/BioDSL/test_serializer.rb +72 -0
- data/test/BioDSL/test_stream.rb +55 -0
- data/test/BioDSL/test_taxonomy.rb +336 -0
- data/test/BioDSL/test_test.rb +42 -0
- data/test/BioDSL/test_tmp_dir.rb +58 -0
- data/test/BioDSL/test_usearch.rb +33 -0
- data/test/BioDSL/test_verbose.rb +42 -0
- data/test/helper.rb +82 -0
- data/www/command.html.haml +14 -0
- data/www/css.html.haml +55 -0
- data/www/input_files.html.haml +3 -0
- data/www/layout.html.haml +12 -0
- data/www/output_files.html.haml +3 -0
- data/www/overview.html.haml +15 -0
- data/www/pipeline.html.haml +4 -0
- data/www/png.html.haml +2 -0
- data/www/status.html.haml +9 -0
- data/www/time.html.haml +11 -0
- metadata +503 -0
@@ -0,0 +1,414 @@
|
|
1
|
+
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< #
|
2
|
+
# #
|
3
|
+
# Copyright (C) 2007-2015 Martin Asser Hansen (mail@maasha.dk). #
|
4
|
+
# #
|
5
|
+
# This program is free software; you can redistribute it and/or #
|
6
|
+
# modify it under the terms of the GNU General Public License #
|
7
|
+
# as published by the Free Software Foundation; either version 2 #
|
8
|
+
# of the License, or (at your option) any later version. #
|
9
|
+
# #
|
10
|
+
# This program is distributed in the hope that it will be useful, #
|
11
|
+
# but WITHOUT ANY WARRANTY; without even the implied warranty of #
|
12
|
+
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the #
|
13
|
+
# GNU General Public License for more details. #
|
14
|
+
# #
|
15
|
+
# You should have received a copy of the GNU General Public License #
|
16
|
+
# along with this program; if not, write to the Free Software #
|
17
|
+
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, #
|
18
|
+
# USA. #
|
19
|
+
# #
|
20
|
+
# http://www.gnu.org/copyleft/gpl.html #
|
21
|
+
# #
|
22
|
+
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< #
|
23
|
+
# #
|
24
|
+
# This software is part of BioDSL (www.github.com/maasha/BioDSL). #
|
25
|
+
# #
|
26
|
+
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< #
|
27
|
+
|
28
|
+
module BioDSL
|
29
|
+
# == Read FASTQ entries from one or more files.
|
30
|
+
#
|
31
|
+
# +read_fastq+ read in sequence entries from FASTQ files. Each sequence entry
|
32
|
+
# consists of a sequence name prefixed by a '>' followed by the sequence name
|
33
|
+
# on a line of its own, followed by one or my lines of sequence until the next
|
34
|
+
# entry or the end of the file. The resulting Biopiece record consists of the
|
35
|
+
# following record type:
|
36
|
+
#
|
37
|
+
# {:SEQ_NAME=>"test",
|
38
|
+
# :SEQ=>"AGCATCGACTAGCAGCATTT",
|
39
|
+
# :SEQ_LEN=>20}
|
40
|
+
#
|
41
|
+
# It is possible to read in pair-end data interleaved by using the +input2+
|
42
|
+
# option. Thus a read is in turn from input and input2. If the
|
43
|
+
# +reverse_complement+ option is used, then the input2 reads will be
|
44
|
+
# reverse-complemented.
|
45
|
+
#
|
46
|
+
# Input files may be compressed with gzip og bzip2.
|
47
|
+
#
|
48
|
+
# For more about the FASTQ format:
|
49
|
+
#
|
50
|
+
# http://en.wikipedia.org/wiki/Fasta_format
|
51
|
+
#
|
52
|
+
# == Usage
|
53
|
+
# read_fastq(input: <glob>[, input2: <glob>[, first: <uint>|last: <uint>
|
54
|
+
# [, reverse_complement: <bool>]]])
|
55
|
+
#
|
56
|
+
# === Options
|
57
|
+
# * input <glob> - Input file or file glob expression.
|
58
|
+
# * input2 <glob> - Input file or file glob expression.
|
59
|
+
# * first <uint> - Only read in the _first_ number of entries.
|
60
|
+
# * last <uint> - Only read in the _last_ number of entries.
|
61
|
+
# * reverse_complement: <bool> - Reverse-complements input2 reads.
|
62
|
+
#
|
63
|
+
# == Examples
|
64
|
+
#
|
65
|
+
# To read all FASTQ entries from a file:
|
66
|
+
#
|
67
|
+
# BP.new.read_fastq(input: "test.fq").dump.run
|
68
|
+
#
|
69
|
+
# To read all FASTQ entries from a gzipped file:
|
70
|
+
#
|
71
|
+
# BP.new.read_fastq(input: "test.fq.gz").dump.run
|
72
|
+
#
|
73
|
+
# To read in only 10 records from a FASTQ file:
|
74
|
+
#
|
75
|
+
# BP.new.read_fastq(input: "test.fq", first: 10).dump.run
|
76
|
+
#
|
77
|
+
# To read in the last 10 records from a FASTQ file:
|
78
|
+
#
|
79
|
+
# BP.new.read_fastq(input: "test.fq", last: 10).dump.run
|
80
|
+
#
|
81
|
+
# To read all FASTQ entries from multiple files:
|
82
|
+
#
|
83
|
+
# BP.new.read_fastq(input: "test1.fq,test2.fq").dump.run
|
84
|
+
#
|
85
|
+
# To read FASTQ entries from multiple files using a glob expression:
|
86
|
+
#
|
87
|
+
# BP.new.read_fastq(input: "*.fq").dump.run
|
88
|
+
#
|
89
|
+
# To read FASTQ entries from pair-end data:
|
90
|
+
#
|
91
|
+
# BP.new.read_fastq(input: "file1.fq", input2: "file2.fq").dump.run
|
92
|
+
#
|
93
|
+
# To read FASTQ entries from pair-end data:
|
94
|
+
#
|
95
|
+
# BP.new.read_fastq(input: "file1.fq", input2: "file2.fq").dump.run
|
96
|
+
#
|
97
|
+
# To read FASTQ entries from pair-end data and reverse-complement read2:
|
98
|
+
#
|
99
|
+
# BP.new.
|
100
|
+
# read_fastq(input: "file1.fq", input2: "file2.fq",
|
101
|
+
# reverse_complement: true)
|
102
|
+
# .dump.run
|
103
|
+
#
|
104
|
+
# rubocop: disable ClassLength
|
105
|
+
# rubocop: disable Metrics/AbcSize
|
106
|
+
# rubocop: disable Metrics/CyclomaticComplexity
|
107
|
+
# rubocop: disable Metrics/PerceivedComplexity
|
108
|
+
class ReadFastq
|
109
|
+
MAX_TEST = 1_000
|
110
|
+
STATS = %i(records_in records_out sequences_in sequences_out residues_in
|
111
|
+
residues_out)
|
112
|
+
|
113
|
+
# Constructor for ReadFastq.
|
114
|
+
#
|
115
|
+
# @param options [Hash] Options hash.
|
116
|
+
# @option options [Symbol,String] :encoding
|
117
|
+
# @option options [String] :input
|
118
|
+
# @option options [String] :input2
|
119
|
+
# @option options [Integer] :first
|
120
|
+
# @option options [Integer] :last
|
121
|
+
# @option options [Boolean] :reverse_complement
|
122
|
+
#
|
123
|
+
# @return [ReadFastq] Class instance.
|
124
|
+
def initialize(options)
|
125
|
+
@options = options
|
126
|
+
@encoding = options[:encoding] ? options[:encoding].to_sym : :auto
|
127
|
+
@pair = options[:input2]
|
128
|
+
@buffer = []
|
129
|
+
@type = nil
|
130
|
+
|
131
|
+
check_options
|
132
|
+
end
|
133
|
+
|
134
|
+
# Return command lambda for ReadFastq.
|
135
|
+
#
|
136
|
+
# @return [Proc] Command lambda.
|
137
|
+
def lmb
|
138
|
+
lambda do |input, output, status|
|
139
|
+
status_init(status, STATS)
|
140
|
+
|
141
|
+
process_input(input, output)
|
142
|
+
|
143
|
+
case
|
144
|
+
when @options[:first] && @pair then read_first_pair(output)
|
145
|
+
when @options[:first] then read_first_single(output)
|
146
|
+
when @options[:last] && @pair then read_last_pair(output)
|
147
|
+
when @options[:last] then read_last_single(output)
|
148
|
+
when @pair then read_all_pair(output)
|
149
|
+
else
|
150
|
+
read_all_single(output)
|
151
|
+
end
|
152
|
+
end
|
153
|
+
end
|
154
|
+
|
155
|
+
private
|
156
|
+
|
157
|
+
# Check options.
|
158
|
+
def check_options
|
159
|
+
options_allowed(@options, :encoding, :input, :input2, :first, :last,
|
160
|
+
:reverse_complement)
|
161
|
+
options_allowed_values(@options, encoding: [:auto, :base_33, :base_64])
|
162
|
+
options_allowed_values(@options, reverse_complement: [nil, true, false])
|
163
|
+
options_tie(@options, reverse_complement: :input2)
|
164
|
+
options_required(@options, :input)
|
165
|
+
options_files_exist(@options, :input, :input2)
|
166
|
+
options_unique(@options, :first, :last)
|
167
|
+
options_assert(@options, ':first >= 0')
|
168
|
+
options_assert(@options, ':last >= 0')
|
169
|
+
end
|
170
|
+
|
171
|
+
# Emit all records from the input stream to the output stream.
|
172
|
+
#
|
173
|
+
# @param input [Enumerator] Input stream.
|
174
|
+
# @param output [Enumerator::Yielder] Output stream.
|
175
|
+
def process_input(input, output)
|
176
|
+
return unless input
|
177
|
+
|
178
|
+
input.each do |record|
|
179
|
+
@status[:records_in] += 1
|
180
|
+
@status[:records_out] += 1
|
181
|
+
|
182
|
+
if (seq = record[:SEQ])
|
183
|
+
@status[:sequences_in] += 1
|
184
|
+
@status[:residues_in] += seq.length
|
185
|
+
end
|
186
|
+
|
187
|
+
output << record
|
188
|
+
end
|
189
|
+
end
|
190
|
+
|
191
|
+
# Read :first FASTQ entries from single files.
|
192
|
+
#
|
193
|
+
# @param output [Enumerator::Yielder] Output stream.
|
194
|
+
def read_first_single(output)
|
195
|
+
fastq_files.each do |file|
|
196
|
+
BioDSL::Fastq.open(file) do |ios|
|
197
|
+
ios.each do |entry|
|
198
|
+
check_entry(entry)
|
199
|
+
output << entry.to_bp
|
200
|
+
@status[:records_out] += 1
|
201
|
+
@status[:sequences_out] += 1
|
202
|
+
@status[:residues_out] += entry.length
|
203
|
+
return if @status[:sequences_out] >= @options[:first]
|
204
|
+
end
|
205
|
+
end
|
206
|
+
end
|
207
|
+
end
|
208
|
+
|
209
|
+
# Read :first FASTQ entries from paired files interleaved.
|
210
|
+
#
|
211
|
+
# @param output [Enumerator::Yielder] Output stream.
|
212
|
+
#
|
213
|
+
# rubocop: disable MethodLength
|
214
|
+
def read_first_pair(output)
|
215
|
+
fastq_files.each_slice(2) do |file1, file2|
|
216
|
+
BioDSL::Fastq.open(file1) do |ios1|
|
217
|
+
BioDSL::Fastq.open(file2) do |ios2|
|
218
|
+
while (entry1 = ios1.next_entry) && (entry2 = ios2.next_entry)
|
219
|
+
check_entry(entry1, entry2)
|
220
|
+
reverse_complement(entry2) if @options[:reverse_complement]
|
221
|
+
output << entry1.to_bp
|
222
|
+
output << entry2.to_bp
|
223
|
+
@status[:records_out] += 2
|
224
|
+
@status[:sequences_out] += 2
|
225
|
+
@status[:residues_out] += entry1.length + entry2.length
|
226
|
+
return if @status[:sequences_out] >= @options[:first]
|
227
|
+
end
|
228
|
+
end
|
229
|
+
end
|
230
|
+
end
|
231
|
+
end
|
232
|
+
|
233
|
+
# Read :last FASTQ entries from single files.
|
234
|
+
#
|
235
|
+
# @param output [Enumerator::Yielder] Output stream.
|
236
|
+
#
|
237
|
+
# rubocop: enable MethodLength
|
238
|
+
def read_last_single(output)
|
239
|
+
fastq_files.each do |file|
|
240
|
+
BioDSL::Fastq.open(file) do |ios|
|
241
|
+
ios.each do |entry|
|
242
|
+
check_entry(entry)
|
243
|
+
@buffer << entry
|
244
|
+
@buffer.shift if @buffer.size > @options[:last]
|
245
|
+
end
|
246
|
+
end
|
247
|
+
end
|
248
|
+
|
249
|
+
output_buffer(output)
|
250
|
+
end
|
251
|
+
|
252
|
+
# Read :last FASTQ entries from paired files interleaved.
|
253
|
+
#
|
254
|
+
# @param output [Enumerator::Yielder] Output stream.
|
255
|
+
def read_last_pair(output)
|
256
|
+
fastq_files.each_slice(2) do |file1, file2|
|
257
|
+
BioDSL::Fastq.open(file1) do |ios1|
|
258
|
+
BioDSL::Fastq.open(file2) do |ios2|
|
259
|
+
while (entry1 = ios1.next_entry) && (entry2 = ios2.next_entry)
|
260
|
+
check_entry(entry1, entry2)
|
261
|
+
reverse_complement(entry2) if @options[:reverse_complement]
|
262
|
+
@buffer << entry1
|
263
|
+
@buffer << entry2
|
264
|
+
@buffer.shift(@buffer.size - @options[:last])
|
265
|
+
end
|
266
|
+
end
|
267
|
+
end
|
268
|
+
end
|
269
|
+
|
270
|
+
output_buffer(output)
|
271
|
+
end
|
272
|
+
|
273
|
+
# Read all FASTQ entries from single files.
|
274
|
+
#
|
275
|
+
# @param output [Enumerator::Yielder] Output stream.
|
276
|
+
def read_all_single(output)
|
277
|
+
fastq_files.each do |file|
|
278
|
+
BioDSL::Fastq.open(file) do |ios|
|
279
|
+
ios.each do |entry|
|
280
|
+
check_entry(entry)
|
281
|
+
output << entry.to_bp
|
282
|
+
@status[:records_out] += 1
|
283
|
+
@status[:sequences_out] += 1
|
284
|
+
@status[:residues_out] += entry.length
|
285
|
+
end
|
286
|
+
end
|
287
|
+
end
|
288
|
+
end
|
289
|
+
|
290
|
+
# Read all FASTQ entries from paired files interleaved.
|
291
|
+
#
|
292
|
+
# @param output [Enumerator::Yielder] Output stream.
|
293
|
+
def read_all_pair(output)
|
294
|
+
fastq_files.each_slice(2) do |file1, file2|
|
295
|
+
BioDSL::Fastq.open(file1) do |ios1|
|
296
|
+
BioDSL::Fastq.open(file2) do |ios2|
|
297
|
+
while (entry1 = ios1.next_entry) && (entry2 = ios2.next_entry)
|
298
|
+
check_entry(entry1, entry2)
|
299
|
+
reverse_complement(entry2) if @options[:reverse_complement]
|
300
|
+
output << entry1.to_bp
|
301
|
+
output << entry2.to_bp
|
302
|
+
@status[:records_out] += 2
|
303
|
+
@status[:sequences_out] += 2
|
304
|
+
@status[:residues_out] += entry1.length + entry2.length
|
305
|
+
end
|
306
|
+
end
|
307
|
+
end
|
308
|
+
end
|
309
|
+
end
|
310
|
+
|
311
|
+
# Return a list of input files or an interleaved list of input files if
|
312
|
+
# :input2 is specified.
|
313
|
+
#
|
314
|
+
# @return [Array] List of FASTQ files.
|
315
|
+
def fastq_files
|
316
|
+
if @options[:input2]
|
317
|
+
files1 = options_glob(@options[:input])
|
318
|
+
files2 = options_glob(@options[:input2])
|
319
|
+
|
320
|
+
check_input_files(files1, files2)
|
321
|
+
|
322
|
+
files1.zip(files2).flatten
|
323
|
+
else
|
324
|
+
options_glob(@options[:input])
|
325
|
+
end
|
326
|
+
end
|
327
|
+
|
328
|
+
# Do the following for the given entry:
|
329
|
+
#
|
330
|
+
# * determine encoding.
|
331
|
+
# * reverse complement if indicated.
|
332
|
+
# * convert encoding
|
333
|
+
# * coerce encoding
|
334
|
+
# * check score range
|
335
|
+
#
|
336
|
+
# @param entries [Array] Sequence entries.
|
337
|
+
def check_entry(*entries)
|
338
|
+
entries.each do |entry|
|
339
|
+
determine_encoding(entry)
|
340
|
+
|
341
|
+
entry.qual_convert!(@encoding, :base_33)
|
342
|
+
entry.qual_coerce!(:base_33)
|
343
|
+
|
344
|
+
check_score_range(entry)
|
345
|
+
end
|
346
|
+
end
|
347
|
+
|
348
|
+
# Reverse complement sequence.
|
349
|
+
#
|
350
|
+
# @param entry [BioDSL::Seq] Sequence entry.
|
351
|
+
def reverse_complement(entry)
|
352
|
+
@type = entry.type_guess unless @type
|
353
|
+
entry.type = @type
|
354
|
+
entry.reverse!.complement!
|
355
|
+
end
|
356
|
+
|
357
|
+
# Check that files1 and files2 are equal.
|
358
|
+
#
|
359
|
+
# @param files1 [Array] List of files.
|
360
|
+
# @param files2 [Array] List of files.
|
361
|
+
#
|
362
|
+
# @raise [BioDSL::OptionError] If not equal.
|
363
|
+
def check_input_files(files1, files2)
|
364
|
+
size1 = files1.size
|
365
|
+
size2 = files2.size
|
366
|
+
return if size1 == size2
|
367
|
+
|
368
|
+
msg = "input and input2 file count don't match: #{size1} != #{size2}"
|
369
|
+
fail BioDSL::OptionError, msg
|
370
|
+
end
|
371
|
+
|
372
|
+
# Check the score range for a given entry.
|
373
|
+
#
|
374
|
+
# @param entry [BioDSL::Seq] Sequence entry.
|
375
|
+
#
|
376
|
+
# @raise [BioDSL::SeqError] If quality score is outside range.
|
377
|
+
def check_score_range(entry)
|
378
|
+
return if @status[:sequences_out] >= MAX_TEST
|
379
|
+
return if entry.qual_valid?(:base_33)
|
380
|
+
fail BioDSL::SeqError, 'Quality score outside valid range'
|
381
|
+
end
|
382
|
+
|
383
|
+
# Determine the quality score encoding.
|
384
|
+
#
|
385
|
+
# @raise [BioDSL::SeqError] If encoding wasn't determined.
|
386
|
+
def determine_encoding(entry)
|
387
|
+
return unless @encoding == :auto
|
388
|
+
|
389
|
+
@encoding = if entry.qual_base33?
|
390
|
+
:base_33
|
391
|
+
elsif entry.qual_base64?
|
392
|
+
:base_64
|
393
|
+
else
|
394
|
+
msg = 'Could not auto-detect quality score encoding'
|
395
|
+
fail BioDSL::SeqError, msg
|
396
|
+
end
|
397
|
+
end
|
398
|
+
|
399
|
+
# Emit all records in the buffer to the output stream.
|
400
|
+
#
|
401
|
+
# @param output [Enumerator::Yielder] Output stream.
|
402
|
+
def output_buffer(output)
|
403
|
+
return unless @options[:last]
|
404
|
+
|
405
|
+
@buffer.each do |entry|
|
406
|
+
output << entry.to_bp
|
407
|
+
|
408
|
+
@status[:records_out] += 1
|
409
|
+
@status[:sequences_out] += 1
|
410
|
+
@status[:residues_out] += entry.length
|
411
|
+
end
|
412
|
+
end
|
413
|
+
end
|
414
|
+
end
|