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 +1,233 @@
|
|
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 the BioDSL framework (www.BioDSL.org). #
|
25
|
+
# #
|
26
|
+
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< #
|
27
|
+
|
28
|
+
module BioDSL
|
29
|
+
# == Plot tabular numerical data in a heatmap.
|
30
|
+
#
|
31
|
+
# A heatmap can be plotted with +plot_heatmap+ using numerical data (Non-
|
32
|
+
# numerical data is ignored). Data should be tabular with records as rows and
|
33
|
+
# keys as columns - the data cells plotted will be the values.
|
34
|
+
#
|
35
|
+
# Default graphics are crufty ASCII and you probably want high resolution
|
36
|
+
# postscript or SVG output instead with is easy using the +terminal+ option.
|
37
|
+
# Plotting is done using GNUplot which allows for different types of output.
|
38
|
+
#
|
39
|
+
# GNUplot must be installed for +plot_heatmap+ to work. Read more here:
|
40
|
+
#
|
41
|
+
# http://www.gnuplot.info/
|
42
|
+
#
|
43
|
+
# == Usage
|
44
|
+
#
|
45
|
+
# plot_heatmap([keys: <list> | skip: <list>[, output: <file>
|
46
|
+
# [, force: <bool> [, terminal: <string>
|
47
|
+
# [, title: <string>[, xlabel: <string>[, ylabel: <string>
|
48
|
+
# [, test: <bool>]]]]]]])
|
49
|
+
#
|
50
|
+
# === Options
|
51
|
+
#
|
52
|
+
# * keys: <list> - Comma separated list of keys to plot as columns.
|
53
|
+
# * skip: <list> - Comma separated list of keys to skip as columns.
|
54
|
+
# * output: <file> - Output file.
|
55
|
+
# * force: <bool> - Force overwrite existing output file.
|
56
|
+
# * terminal: <string> - Terminal for output: dumb|post|svg|x11|aqua|png|pdf
|
57
|
+
# (default=dumb).
|
58
|
+
# * title: <string> - Plot title (default="Heatmap").
|
59
|
+
# * xlabel: <string> - X-axis label (default="x").
|
60
|
+
# * ylabel: <string> - Y-axis label (default="y").
|
61
|
+
# * test: <bool> - Output Gnuplot script instead of plot.
|
62
|
+
#
|
63
|
+
# == Examples
|
64
|
+
#
|
65
|
+
# Here we plot a heatmap of data a table:
|
66
|
+
#
|
67
|
+
# BP.new.read_table(input: "test.tab").plot_heatmap.run
|
68
|
+
#
|
69
|
+
# rubocop:disable ClassLength
|
70
|
+
class PlotHeatmap
|
71
|
+
require 'gnuplotter'
|
72
|
+
require 'set'
|
73
|
+
require 'BioDSL/helpers/aux_helper'
|
74
|
+
|
75
|
+
include AuxHelper
|
76
|
+
|
77
|
+
STATS = %i(records_in records_out)
|
78
|
+
|
79
|
+
# Constructor for PlotHeatmap.
|
80
|
+
#
|
81
|
+
# @param options [Hash] Options hash.
|
82
|
+
# @option options [Array] :keys List of keys to plot as column.
|
83
|
+
# @option options [Array] :skip List of keys to skip as column.
|
84
|
+
# @option options [String] :output Path to output file.
|
85
|
+
# @option options [Boolean] :forcea Flag to force overwrite output file.
|
86
|
+
# @option options [Symbol] :terminal Set plot terminal type.
|
87
|
+
# @option options [String] :title Set plot title.
|
88
|
+
# @option options [String] :xlabel Set plot xlabel.
|
89
|
+
# @option options [String] :ylabel Set plot ylabel
|
90
|
+
# @option options [Boolean] :logscale Logscale Z-axis.
|
91
|
+
# @option options [Boolean] :test Output gnuplot script.
|
92
|
+
#
|
93
|
+
# @return [PlotHeatmap] Class instance.
|
94
|
+
def initialize(options)
|
95
|
+
@options = options
|
96
|
+
@headings = nil
|
97
|
+
@skip_keys = determine_skip_keys
|
98
|
+
|
99
|
+
aux_exist('gnuplot')
|
100
|
+
check_options
|
101
|
+
defaults
|
102
|
+
end
|
103
|
+
|
104
|
+
# Return command lambda for plot_histogram.
|
105
|
+
#
|
106
|
+
# @return [Proc] Command lambda.
|
107
|
+
def lmb
|
108
|
+
lambda do |input, output, status|
|
109
|
+
status_init(status, STATS)
|
110
|
+
|
111
|
+
gp = GnuPlotter.new
|
112
|
+
|
113
|
+
plot_options(gp)
|
114
|
+
plot_dataset(gp, input, output)
|
115
|
+
plot_output(gp)
|
116
|
+
end
|
117
|
+
end
|
118
|
+
|
119
|
+
private
|
120
|
+
|
121
|
+
# Check options.
|
122
|
+
def check_options
|
123
|
+
options_allowed(@options, :keys, :skip, :output, :force, :terminal,
|
124
|
+
:title, :xlabel, :ylabel, :logscale, :test)
|
125
|
+
options_unique(@options, :keys, :skip)
|
126
|
+
options_allowed_values(@options, terminal: [:dumb, :post, :svg, :x11,
|
127
|
+
:aqua, :png, :pdf])
|
128
|
+
options_allowed_values(@options, test: [nil, true, false])
|
129
|
+
options_allowed_values(@options, logscale: [nil, true, false])
|
130
|
+
options_files_exist_force(@options, :output)
|
131
|
+
end
|
132
|
+
|
133
|
+
# Set default options.
|
134
|
+
def defaults
|
135
|
+
@options[:terminal] ||= :dumb
|
136
|
+
@options[:title] ||= 'Heatmap'
|
137
|
+
@options[:xlabel] ||= 'x'
|
138
|
+
@options[:ylabel] ||= 'y'
|
139
|
+
end
|
140
|
+
|
141
|
+
# Compile a set of keys to skip.
|
142
|
+
#
|
143
|
+
# @return [Set] Set of keys to skip.
|
144
|
+
def determine_skip_keys
|
145
|
+
return unless @options[:skip]
|
146
|
+
@options[:skip].each_with_object(Set.new) { |e, a| a << e.to_sym }
|
147
|
+
end
|
148
|
+
|
149
|
+
# Determine the headings.
|
150
|
+
#
|
151
|
+
# @param record [Hash] BioDSL record.
|
152
|
+
def determine_headings(record)
|
153
|
+
@headings =
|
154
|
+
if @options[:keys]
|
155
|
+
@options[:keys].map(&:to_sym)
|
156
|
+
elsif record.keys.first =~ /^V\d+$/
|
157
|
+
sort_keys(record)
|
158
|
+
else
|
159
|
+
record.keys
|
160
|
+
end
|
161
|
+
|
162
|
+
@headings.reject! { |r| @skip_keys.include? r } if @options[:skip]
|
163
|
+
end
|
164
|
+
|
165
|
+
# Sort records keys numerically, when the keys are in the format Vn, where n
|
166
|
+
# is an Integer.
|
167
|
+
#
|
168
|
+
# @param record [Hash] BioDSL record.
|
169
|
+
#
|
170
|
+
# @return [Array] List of sorted keys.
|
171
|
+
def sort_keys(record)
|
172
|
+
record.keys.sort do |a, b|
|
173
|
+
a.to_s[1..a.to_s.size].to_i <=> b.to_s[1..a.to_s.size].to_i
|
174
|
+
end
|
175
|
+
end
|
176
|
+
|
177
|
+
# Set options for plot.
|
178
|
+
#
|
179
|
+
# @param gp [GnuPlotter] GnuPlotter object.
|
180
|
+
def plot_options(gp)
|
181
|
+
gp.set terminal: @options[:terminal].to_s
|
182
|
+
gp.set title: @options[:title]
|
183
|
+
gp.set xlabel: @options[:xlabel]
|
184
|
+
gp.set ylabel: @options[:ylabel]
|
185
|
+
gp.set output: @options[:output] if @options[:output]
|
186
|
+
gp.set view: 'map'
|
187
|
+
gp.set autoscale: 'xfix'
|
188
|
+
gp.set autoscale: 'yfix'
|
189
|
+
gp.set nokey: true
|
190
|
+
gp.set tic: 'scale 0'
|
191
|
+
gp.set palette: 'rgbformulae 22,13,10'
|
192
|
+
gp.set logscale: 'cb' if @options[:logscale]
|
193
|
+
gp.unset xtics: true
|
194
|
+
gp.unset ytics: true
|
195
|
+
end
|
196
|
+
|
197
|
+
# Plot relevant data from the input stream.
|
198
|
+
#
|
199
|
+
# @param gp [GnuPlotter] GnuPlotter object.
|
200
|
+
# @param input [Enumerator] Input stream.
|
201
|
+
# @param output [Enumerator::Yielder] Output stream.
|
202
|
+
def plot_dataset(gp, input, output)
|
203
|
+
gp.add_dataset(matrix: :true, with: 'image') do |plotter|
|
204
|
+
input.each do |record|
|
205
|
+
@status[:records_in] += 1
|
206
|
+
|
207
|
+
determine_headings(record) unless @headings
|
208
|
+
|
209
|
+
plotter << record.values_at(*@headings)
|
210
|
+
|
211
|
+
next unless output
|
212
|
+
|
213
|
+
output << record
|
214
|
+
|
215
|
+
@status[:records_out] += 1
|
216
|
+
end
|
217
|
+
end
|
218
|
+
end
|
219
|
+
|
220
|
+
# Output plot data according to options.
|
221
|
+
#
|
222
|
+
# @param gp [GnuPlotter] GnuPlotter object.
|
223
|
+
def plot_output(gp)
|
224
|
+
if @options[:test]
|
225
|
+
$stderr.puts gp.to_gp
|
226
|
+
elsif @options[:terminal] == :dumb
|
227
|
+
puts gp.splot
|
228
|
+
else
|
229
|
+
gp.splot
|
230
|
+
end
|
231
|
+
end
|
232
|
+
end
|
233
|
+
end
|
@@ -0,0 +1,306 @@
|
|
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 the BioDSL framework (www.BioDSL.org). #
|
25
|
+
# #
|
26
|
+
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< #
|
27
|
+
|
28
|
+
# rubocop:disable LineLength
|
29
|
+
module BioDSL
|
30
|
+
# == Plot a histogram of numerical values for a specified key.
|
31
|
+
#
|
32
|
+
# +plot_histogram+ create a histogram plot of the values for a specified key
|
33
|
+
# from all records in the stream. Plotting is done using GNUplot which allows
|
34
|
+
# for different types of output the default one being crufty ASCII graphics.
|
35
|
+
#
|
36
|
+
# GNUplot's facility for setting the xrange labels is used for numeric values,
|
37
|
+
# while for non-numeric values these are used for xrange labels.
|
38
|
+
#
|
39
|
+
# GNUplot must be installed for plot_histogram to work. Read more here:
|
40
|
+
#
|
41
|
+
# http://www.gnuplot.info/
|
42
|
+
#
|
43
|
+
# == Usage
|
44
|
+
#
|
45
|
+
# plot_histogram(<key: <string>>[, value: <string>[, output: <file>
|
46
|
+
# [, force: <bool>[, terminal: <string>[, title: <string>
|
47
|
+
# [, xlabel: <string>[, ylabel: <string>
|
48
|
+
# [, ylogscale: <bool>[, test: <bool>]]]]]]]]])
|
49
|
+
#
|
50
|
+
# === Options
|
51
|
+
#
|
52
|
+
# * key: <string> - Key to use for plotting.
|
53
|
+
# * value: <string> - Alternative key who's value to use.
|
54
|
+
# * output: <file> - Output file.
|
55
|
+
# * force: <bool> - Force overwrite existing output file.
|
56
|
+
# * terminal: <string> - Terminal for output: dumb|post|svg|x11|aqua|png|pdf
|
57
|
+
# (default=dumb).
|
58
|
+
# * title: <string> - Plot title (default="Histogram").
|
59
|
+
# * xlabel: <string> - X-axis label (default=<key>).
|
60
|
+
# * ylabel: <string> - Y-axis label (default="n").
|
61
|
+
# * ylogscale: <bool> - Set y-axis to log scale.
|
62
|
+
# * test: <bool> - Output Gnuplot script instead of plot.
|
63
|
+
#
|
64
|
+
# == Examples
|
65
|
+
#
|
66
|
+
# Here we plot a histogram of sequence lengths from a FASTA file:
|
67
|
+
#
|
68
|
+
# read_fasta(input: "test.fna").plot_histogram(key: :SEQ_LEN).run
|
69
|
+
#
|
70
|
+
# Histogram
|
71
|
+
# + + + + + +
|
72
|
+
# 90 +++-------------+------------+------------+------------+-------------+++
|
73
|
+
# | |
|
74
|
+
# 80 ++ **++
|
75
|
+
# | **|
|
76
|
+
# 70 ++ **++
|
77
|
+
# 60 ++ **++
|
78
|
+
# | **|
|
79
|
+
# 50 ++ **++
|
80
|
+
# | **|
|
81
|
+
# 40 ++ **++
|
82
|
+
# | **|
|
83
|
+
# 30 ++ **++
|
84
|
+
# 20 ++ **++
|
85
|
+
# | **|
|
86
|
+
# 10 ++ **++
|
87
|
+
# | ******|
|
88
|
+
# 0 +++-------------+------------+**--------**+--***-------+**--**********++
|
89
|
+
# + + + + + +
|
90
|
+
# 0 10 20 30 40 50
|
91
|
+
# SEQ_LEN
|
92
|
+
#
|
93
|
+
# To render X11 output (i.e. instant view) use the +terminal+ option:
|
94
|
+
#
|
95
|
+
# read_fasta(input: "test.fna").
|
96
|
+
# plot_histogram(key: :SEQ_LEN, terminal: :x11).run
|
97
|
+
#
|
98
|
+
# To generate a PNG image and save to file:
|
99
|
+
#
|
100
|
+
# read_fasta(input: "test.fna").
|
101
|
+
# plot_histogram(key: :SEQ_LEN, terminal: :png, output: "plot.png").run
|
102
|
+
#
|
103
|
+
# rubocop:disable ClassLength
|
104
|
+
# rubocop:enable LineLength
|
105
|
+
class PlotHistogram
|
106
|
+
require 'gnuplotter'
|
107
|
+
require 'BioDSL/helpers/aux_helper'
|
108
|
+
|
109
|
+
include AuxHelper
|
110
|
+
|
111
|
+
STATS = %i(records_in records_out)
|
112
|
+
|
113
|
+
# Constructor for PlotHistogram.
|
114
|
+
#
|
115
|
+
# @param options [Hash] Options hash.
|
116
|
+
# @option options [String,:Symbol] :key
|
117
|
+
# @option options [String,:Symbol] :value
|
118
|
+
# @option options [String] :output
|
119
|
+
# @option options [Booleon] :force
|
120
|
+
# @option options [String,:Symbol] :terminal
|
121
|
+
# @option options [String] :title
|
122
|
+
# @option options [String] :xlabel
|
123
|
+
# @option options [String] :ylabel
|
124
|
+
# @option options [Booleon] :ylogscale
|
125
|
+
# @option options [Booleon] :test
|
126
|
+
#
|
127
|
+
# @return [PlotHistogram] class instance.
|
128
|
+
def initialize(options)
|
129
|
+
@options = options
|
130
|
+
@key = options[:key]
|
131
|
+
@value = options[:value]
|
132
|
+
@count_hash = Hash.new(0)
|
133
|
+
@gp = nil
|
134
|
+
|
135
|
+
aux_exist('gnuplot')
|
136
|
+
check_options
|
137
|
+
defaults
|
138
|
+
end
|
139
|
+
|
140
|
+
# Return the command lambda for plot_histogram
|
141
|
+
#
|
142
|
+
# @return [Proc] command lambda.
|
143
|
+
def lmb
|
144
|
+
lambda do |input, output, status|
|
145
|
+
status_init(status, STATS)
|
146
|
+
|
147
|
+
process_input(input, output)
|
148
|
+
plot_create
|
149
|
+
plot_output
|
150
|
+
end
|
151
|
+
end
|
152
|
+
|
153
|
+
private
|
154
|
+
|
155
|
+
# Check options.
|
156
|
+
def check_options
|
157
|
+
options_allowed(@options, :key, :value, :output, :force, :terminal,
|
158
|
+
:title, :xlabel, :ylabel, :ylogscale, :test)
|
159
|
+
options_allowed_values(@options, terminal: [:dumb, :post, :svg, :x11,
|
160
|
+
:aqua, :png, :pdf])
|
161
|
+
options_allowed_values(@options, force: [nil, true, false])
|
162
|
+
options_allowed_values(@options, test: [nil, true, false])
|
163
|
+
options_required(@options, :key)
|
164
|
+
options_files_exist_force(@options, :output)
|
165
|
+
end
|
166
|
+
|
167
|
+
# Set default values for options hash.
|
168
|
+
def defaults
|
169
|
+
@options[:terminal] ||= :dumb
|
170
|
+
@options[:title] ||= 'Histogram'
|
171
|
+
@options[:xlabel] ||= @options[:key]
|
172
|
+
@options[:ylabel] ||= 'n'
|
173
|
+
|
174
|
+
@options[:ylogscale] &&
|
175
|
+
@options[:ylabel] = "log10(#{@options[:ylabel]})"
|
176
|
+
end
|
177
|
+
|
178
|
+
# Process the input stream, collect all plot data, and output records.
|
179
|
+
#
|
180
|
+
# @param input [Enumerator] Input stream.
|
181
|
+
# @param output [Enumerator::Yielder] Output stream.
|
182
|
+
def process_input(input, output)
|
183
|
+
input.each do |record|
|
184
|
+
@status[:records_in] += 1
|
185
|
+
|
186
|
+
if (k = record[@key])
|
187
|
+
if @value
|
188
|
+
if (v = record[@value])
|
189
|
+
@count_hash[k] += v
|
190
|
+
else
|
191
|
+
fail "value: #{@value} not found in record: #{record}"
|
192
|
+
end
|
193
|
+
else
|
194
|
+
@count_hash[k] += 1
|
195
|
+
end
|
196
|
+
end
|
197
|
+
|
198
|
+
process_output(output, record)
|
199
|
+
end
|
200
|
+
end
|
201
|
+
|
202
|
+
# Output record to the output stream if such is defined.
|
203
|
+
#
|
204
|
+
# @param output [Enumerator::Yielder] Output stream.
|
205
|
+
# @param record [Hash] BioDSL record.
|
206
|
+
def process_output(output, record)
|
207
|
+
return unless output
|
208
|
+
output << record
|
209
|
+
@status[:records_out] += 1
|
210
|
+
end
|
211
|
+
|
212
|
+
# Create a Gnuplot using the collected data from the input stream.
|
213
|
+
def plot_create
|
214
|
+
@gp = GnuPlotter.new
|
215
|
+
plot_defaults
|
216
|
+
plot_fix_ylogscale
|
217
|
+
|
218
|
+
if @count_hash.empty?
|
219
|
+
plot_empty
|
220
|
+
elsif @count_hash.keys.first.is_a? Numeric
|
221
|
+
plot_numeric
|
222
|
+
else
|
223
|
+
plot_string
|
224
|
+
end
|
225
|
+
|
226
|
+
plot_fix_xtics
|
227
|
+
end
|
228
|
+
|
229
|
+
# Set the default values for the plot.
|
230
|
+
def plot_defaults
|
231
|
+
@gp.set terminal: @options[:terminal].to_s
|
232
|
+
@gp.set title: @options[:title]
|
233
|
+
@gp.set xlabel: @options[:xlabel]
|
234
|
+
@gp.set ylabel: @options[:ylabel]
|
235
|
+
@gp.set autoscale: 'xfix'
|
236
|
+
@gp.set style: 'fill solid 0.5 border'
|
237
|
+
@gp.set xtics: 'out'
|
238
|
+
@gp.set ytics: 'out'
|
239
|
+
end
|
240
|
+
|
241
|
+
# Set plot values accodingly if the ylogscale flag is set.
|
242
|
+
def plot_fix_ylogscale
|
243
|
+
if @options[:ylogscale]
|
244
|
+
@gp.set logscale: 'y'
|
245
|
+
@gp.set yrange: '[1:*]'
|
246
|
+
else
|
247
|
+
@gp.set yrange: '[0:*]'
|
248
|
+
end
|
249
|
+
end
|
250
|
+
|
251
|
+
# Set plot values to create an empty plot if no plot data was collected.
|
252
|
+
def plot_empty
|
253
|
+
@gp.set yrange: '[-1:1]'
|
254
|
+
@gp.set key: 'off'
|
255
|
+
@gp.unset xtics: true
|
256
|
+
@gp.unset ytics: true
|
257
|
+
end
|
258
|
+
|
259
|
+
# If plot data have numeric xtic values use numeric xtic labels.
|
260
|
+
def plot_numeric
|
261
|
+
x_max = @count_hash.keys.max || 0
|
262
|
+
|
263
|
+
@gp.add_dataset(using: '1:2', with: 'boxes notitle') do |plotter|
|
264
|
+
(0..x_max).each { |x| plotter << [x, @count_hash[x]] }
|
265
|
+
end
|
266
|
+
end
|
267
|
+
|
268
|
+
# If plot data gave string xtic values use these as xtic labels.
|
269
|
+
def plot_string
|
270
|
+
plot_xtics_rotate
|
271
|
+
|
272
|
+
@gp.add_dataset(using: '2:xticlabels(1)',
|
273
|
+
with: 'boxes notitle lc rgb "red"') do |plotter|
|
274
|
+
@count_hash.each { |k, v| plotter << [k, v] }
|
275
|
+
end
|
276
|
+
end
|
277
|
+
|
278
|
+
# If xtic labels are longer then 2, rotate these.
|
279
|
+
def plot_xtics_rotate
|
280
|
+
return unless @count_hash.first.first.size > 2
|
281
|
+
@gp.set xtics: 'rotate'
|
282
|
+
@gp.set xlabel: ''
|
283
|
+
end
|
284
|
+
|
285
|
+
# Determine if xtics should be plottet and unset these if not. Don't plot
|
286
|
+
# xtics if more than 50 strings.
|
287
|
+
def plot_fix_xtics
|
288
|
+
return unless @count_hash.keys.first.class == String &&
|
289
|
+
@count_hash.size > 50
|
290
|
+
@gp.unset xtics: true
|
291
|
+
end
|
292
|
+
|
293
|
+
# Output plot data
|
294
|
+
def plot_output
|
295
|
+
@gp.set output: @options[:output] if @options[:output]
|
296
|
+
|
297
|
+
if @options[:test]
|
298
|
+
$stderr.puts @gp.to_gp
|
299
|
+
elsif @options[:terminal] == :dumb
|
300
|
+
puts @gp.plot
|
301
|
+
else
|
302
|
+
@gp.plot
|
303
|
+
end
|
304
|
+
end
|
305
|
+
end
|
306
|
+
end
|