bio-samtools 0.6.2 → 2.0.3
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/Gemfile +3 -2
- data/README.md +4 -7
- data/VERSION +1 -1
- data/bio-samtools.gemspec +47 -105
- data/doc/Bio.html +68 -131
- data/doc/Bio/DB.html +51 -111
- data/doc/Bio/DB/Alignment.html +135 -363
- data/doc/Bio/DB/Pileup.html +183 -170
- data/doc/Bio/DB/SAM.html +1396 -820
- data/doc/Bio/DB/SAM/Library.html +73 -123
- data/doc/Bio/DB/SAM/Tools.html +51 -273
- data/doc/Bio/DB/Tag.html +78 -124
- data/doc/Bio/DB/Vcf.html +111 -147
- data/doc/LICENSE_txt.html +113 -148
- data/doc/created.rid +9 -10
- data/doc/fonts.css +167 -0
- data/doc/fonts/Lato-Light.ttf +0 -0
- data/doc/fonts/Lato-LightItalic.ttf +0 -0
- data/doc/fonts/Lato-Regular.ttf +0 -0
- data/doc/fonts/Lato-RegularItalic.ttf +0 -0
- data/doc/fonts/SourceCodePro-Bold.ttf +0 -0
- data/doc/fonts/SourceCodePro-Regular.ttf +0 -0
- data/doc/images/add.png +0 -0
- data/doc/images/arrow_up.png +0 -0
- data/doc/images/delete.png +0 -0
- data/doc/images/tag_blue.png +0 -0
- data/doc/index.html +48 -54
- data/doc/js/darkfish.js +9 -22
- data/doc/js/search.js +20 -5
- data/doc/js/search_index.js +1 -1
- data/doc/rdoc.css +255 -218
- data/doc/table_of_contents.html +256 -137
- data/ext/Rakefile +57 -0
- data/lib/bio-samtools.rb +7 -2
- data/lib/bio/BIOExtensions.rb +89 -0
- data/lib/bio/db/alignment.rb +59 -0
- data/lib/bio/db/fastadb.rb +255 -0
- data/lib/bio/db/pileup.rb +221 -172
- data/lib/bio/db/sam.rb +639 -589
- data/lib/bio/db/sam/{faidx.rb → faidx_old.rb} +0 -0
- data/lib/bio/db/vcf.rb +69 -68
- data/test/.gitignore +1 -0
- data/test/{test_basic.rb → old_test_basic.rb} +33 -1
- data/test/samples/small/dupes.bam +0 -0
- data/test/samples/small/dupes.sam +274 -0
- data/test/samples/small/map_for_reheader.sam +8 -0
- data/test/samples/small/map_to_merge1.bam +0 -0
- data/test/samples/small/map_to_merge1.bam.bai +0 -0
- data/test/samples/small/map_to_merge1.sam +8 -0
- data/test/samples/small/map_to_merge2.bam +0 -0
- data/test/samples/small/map_to_merge2.bam.bai +0 -0
- data/test/samples/small/map_to_merge2.sam +8 -0
- data/test/samples/small/no_md.sam +8 -0
- data/test/samples/small/test_chr.fasta.1.bt2 +0 -0
- data/test/samples/small/test_chr.fasta.2.bt2 +0 -0
- data/test/samples/small/test_chr.fasta.3.bt2 +0 -0
- data/test/samples/small/test_chr.fasta.4.bt2 +0 -0
- data/test/samples/small/test_chr.fasta.rev.1.bt2 +0 -0
- data/test/samples/small/test_chr.fasta.rev.2.bt2 +0 -0
- data/test/samples/small/test_cov.svg +273 -0
- data/test/samples/small/testu.bam.bai +0 -0
- data/test/svg +133 -0
- data/test/test_pileup.rb +84 -0
- data/test/test_sam.rb +331 -0
- data/test/test_vcf.rb +11 -0
- data/{doc → tutorial}/tutorial.html +0 -0
- data/{doc → tutorial}/tutorial.pdf +0 -0
- metadata +56 -114
- data/doc/Bio/DB/SAM/Tools/Bam1CoreT.html +0 -159
- data/doc/Bio/DB/SAM/Tools/Bam1T.html +0 -220
- data/doc/Bio/DB/SAM/Tools/BamHeaderT.html +0 -249
- data/doc/Bio/DB/SAM/Tools/BamPileup1T.html +0 -159
- data/doc/Bio/DB/SAM/Tools/SamfileT.html +0 -171
- data/doc/Bio/DB/SAM/Tools/SamfileTX.html +0 -159
- data/doc/Bio/DB/SAMException.html +0 -205
- data/doc/LibC.html +0 -155
- data/doc/Pileup.html +0 -571
- data/doc/Vcf.html +0 -473
- data/doc/basic_styles.css +0 -31
- data/doc/classes/Bio.html +0 -139
- data/doc/classes/Bio/DB.html +0 -137
- data/doc/classes/Bio/DB/Alignment.html +0 -441
- data/doc/classes/Bio/DB/Alignment.src/M000012.html +0 -19
- data/doc/classes/Bio/DB/Alignment.src/M000013.html +0 -27
- data/doc/classes/Bio/DB/Alignment.src/M000014.html +0 -45
- data/doc/classes/Bio/DB/Alignment.src/M000015.html +0 -40
- data/doc/classes/Bio/DB/SAM.html +0 -510
- data/doc/classes/Bio/DB/SAM/Library.html +0 -135
- data/doc/classes/Bio/DB/SAM/Library.src/M000006.html +0 -28
- data/doc/classes/Bio/DB/SAM/Tools.html +0 -278
- data/doc/classes/Bio/DB/SAM/Tools.src/M000007.html +0 -20
- data/doc/classes/Bio/DB/SAM/Tools/Bam1CoreT.html +0 -111
- data/doc/classes/Bio/DB/SAM/Tools/Bam1T.html +0 -150
- data/doc/classes/Bio/DB/SAM/Tools/Bam1T.src/M000010.html +0 -20
- data/doc/classes/Bio/DB/SAM/Tools/BamHeaderT.html +0 -169
- data/doc/classes/Bio/DB/SAM/Tools/BamHeaderT.src/M000008.html +0 -19
- data/doc/classes/Bio/DB/SAM/Tools/BamHeaderT.src/M000009.html +0 -18
- data/doc/classes/Bio/DB/SAM/Tools/BamPileup1T.html +0 -111
- data/doc/classes/Bio/DB/SAM/Tools/SamfileT.html +0 -129
- data/doc/classes/Bio/DB/SAM/Tools/SamfileTX.html +0 -111
- data/doc/classes/Bio/DB/SAMException.html +0 -140
- data/doc/classes/Bio/DB/SAMException.src/M000016.html +0 -18
- data/doc/classes/Bio/DB/Sam.src/M000017.html +0 -43
- data/doc/classes/Bio/DB/Sam.src/M000018.html +0 -42
- data/doc/classes/Bio/DB/Sam.src/M000019.html +0 -18
- data/doc/classes/Bio/DB/Sam.src/M000020.html +0 -22
- data/doc/classes/Bio/DB/Sam.src/M000021.html +0 -19
- data/doc/classes/Bio/DB/Sam.src/M000022.html +0 -25
- data/doc/classes/Bio/DB/Sam.src/M000023.html +0 -28
- data/doc/classes/Bio/DB/Sam.src/M000024.html +0 -28
- data/doc/classes/Bio/DB/Sam.src/M000025.html +0 -46
- data/doc/classes/Bio/DB/Sam.src/M000026.html +0 -24
- data/doc/classes/Bio/DB/Sam.src/M000027.html +0 -19
- data/doc/classes/Bio/DB/Sam.src/M000028.html +0 -24
- data/doc/classes/Bio/DB/Sam.src/M000029.html +0 -41
- data/doc/classes/Bio/DB/Sam.src/M000030.html +0 -31
- data/doc/classes/Bio/DB/Sam.src/M000031.html +0 -86
- data/doc/classes/Bio/DB/Sam.src/M000032.html +0 -34
- data/doc/classes/Bio/DB/Tag.html +0 -160
- data/doc/classes/Bio/DB/Tag.src/M000011.html +0 -21
- data/doc/classes/LibC.html +0 -105
- data/doc/classes/Pileup.html +0 -374
- data/doc/classes/Pileup.src/M000001.html +0 -34
- data/doc/classes/Pileup.src/M000002.html +0 -21
- data/doc/classes/Pileup.src/M000003.html +0 -21
- data/doc/classes/Pileup.src/M000004.html +0 -21
- data/doc/classes/Pileup.src/M000005.html +0 -31
- data/doc/files/lib/bio-samtools_rb.html +0 -109
- data/doc/files/lib/bio/db/sam/bam_rb.html +0 -108
- data/doc/files/lib/bio/db/sam/faidx_rb.html +0 -108
- data/doc/files/lib/bio/db/sam/library_rb.html +0 -101
- data/doc/files/lib/bio/db/sam/pileup_rb.html +0 -178
- data/doc/files/lib/bio/db/sam/sam_rb.html +0 -113
- data/doc/files/lib/bio/db/sam_rb.html +0 -111
- data/doc/fr_class_index.html +0 -43
- data/doc/fr_file_index.html +0 -33
- data/doc/fr_method_index.html +0 -58
- data/doc/lib/bio-samtools_rb.html +0 -115
- data/doc/lib/bio/db/pileup_rb.html +0 -171
- data/doc/lib/bio/db/sam/bam_rb.html +0 -121
- data/doc/lib/bio/db/sam/faidx_rb.html +0 -117
- data/doc/lib/bio/db/sam/library_rb.html +0 -115
- data/doc/lib/bio/db/sam/pileup_rb.html +0 -171
- data/doc/lib/bio/db/sam/sam_rb.html +0 -121
- data/doc/lib/bio/db/sam/vcf_rb.html +0 -124
- data/doc/lib/bio/db/sam_rb.html +0 -115
- data/doc/lib/bio/db/vcf_rb.html +0 -124
- data/doc/rdoc-style.css +0 -208
- data/lib/bio/db/sam/bam.rb +0 -210
- data/lib/bio/db/sam/sam.rb +0 -86
- data/test/samples/pipe_char/test.bam +0 -0
- data/test/samples/pipe_char/test.bam.bai +0 -0
- data/test/samples/pipe_char/test.tam +0 -10
- data/test/samples/pipe_char/test_chr.fasta +0 -1000
- data/test/samples/pipe_char/test_chr.fasta.fai +0 -1
- data/test/samples/small/test +0 -0
- data/test/samples/small/test.bam +0 -0
- data/test/samples/small/test.fa +0 -20
- data/test/samples/small/test.fai +0 -0
data/test/test_pileup.rb
ADDED
@@ -0,0 +1,84 @@
|
|
1
|
+
$: << File.expand_path(File.dirname(__FILE__) + '/../lib')
|
2
|
+
$: << File.expand_path('.')
|
3
|
+
require 'rubygems'
|
4
|
+
require 'bio/db/pileup'
|
5
|
+
require "test/unit"
|
6
|
+
gem 'test-unit'
|
7
|
+
|
8
|
+
|
9
|
+
class TestPileup < Test::Unit::TestCase
|
10
|
+
|
11
|
+
def setup
|
12
|
+
@pu = Bio::DB::Pileup.new("seq1 279 C 23 A..T,,.,.,...,,,.,..... ;75&<<<<<<<<<=<<<9<<:<<")
|
13
|
+
#a snp...
|
14
|
+
@pu2 = Bio::DB::Pileup.new("seq1 279 C 23 ATTT,,.,.TTTT,,,.,TTTTT ;75&<<<<<<<<<=<<<9<<:<<")
|
15
|
+
#an indel..
|
16
|
+
@pu3 = Bio::DB::Pileup.new("seq2 156 * +AG/+AG 71 252 99 11 +AG * 3 8 0")
|
17
|
+
end
|
18
|
+
|
19
|
+
def test_non_ref_count
|
20
|
+
assert_equal(2, @pu.non_ref_count)
|
21
|
+
end
|
22
|
+
|
23
|
+
def test_ref_count
|
24
|
+
assert_equal(21, @pu.ref_count)
|
25
|
+
end
|
26
|
+
|
27
|
+
def test_consensus
|
28
|
+
assert_equal('C', @pu.consensus)
|
29
|
+
assert_equal('T', @pu2.consensus)
|
30
|
+
end
|
31
|
+
|
32
|
+
def test_non_refs
|
33
|
+
assert_equal(1, @pu.non_refs[:T])
|
34
|
+
assert_equal(1, @pu.non_refs[:A])
|
35
|
+
assert_equal(0, @pu.non_refs[:G])
|
36
|
+
assert_equal(0, @pu.non_refs[:C])
|
37
|
+
end
|
38
|
+
|
39
|
+
|
40
|
+
def test_to_vcf
|
41
|
+
@vcf = Bio::DB::Vcf.new(@pu.to_vcf)
|
42
|
+
assert_equal('seq1', @vcf.chrom)
|
43
|
+
end
|
44
|
+
|
45
|
+
|
46
|
+
def test_indel_gt
|
47
|
+
indel = @pu3.send(:indel_gt)
|
48
|
+
assert_equal('IAG', indel[0])
|
49
|
+
assert_equal('1/1', indel[1])
|
50
|
+
end
|
51
|
+
|
52
|
+
def test_snp_gt
|
53
|
+
snp = @pu2.send(:snp_gt)
|
54
|
+
assert_equal('T,', snp[0])
|
55
|
+
assert_equal('1/2', snp[1])
|
56
|
+
end
|
57
|
+
|
58
|
+
def test_genotype_list
|
59
|
+
gt2 = @pu2.genotype_list
|
60
|
+
gt3 = @pu3.genotype_list
|
61
|
+
assert_equal('T,', gt2[0])
|
62
|
+
assert_equal('1/2', gt2[1])
|
63
|
+
assert_equal('IAG', gt3[0])
|
64
|
+
assert_equal('1/1', gt3[1])
|
65
|
+
end
|
66
|
+
|
67
|
+
def test_iupac_to_base
|
68
|
+
iupac = Bio::DB::Pileup.iupac_to_base('R')
|
69
|
+
iupac.each do |pu|
|
70
|
+
assert_send([['A' , 'G'], :member?, pu])
|
71
|
+
end
|
72
|
+
end
|
73
|
+
|
74
|
+
def test_parse_indel
|
75
|
+
assert_equal('IAG/+AG', @pu3.parse_indel(@pu3.consensus))
|
76
|
+
end
|
77
|
+
|
78
|
+
def test_to_s
|
79
|
+
#check whether there are the correct number of tabs in the string (number of columns -1)
|
80
|
+
assert_equal(12, @pu3.to_s.count("\t"))
|
81
|
+
assert_equal(5, @pu.to_s.count("\t"))
|
82
|
+
end
|
83
|
+
|
84
|
+
end
|
data/test/test_sam.rb
ADDED
@@ -0,0 +1,331 @@
|
|
1
|
+
$: << File.expand_path(File.dirname(__FILE__) + '/../lib')
|
2
|
+
$: << File.expand_path('.')
|
3
|
+
require 'rubygems'
|
4
|
+
require 'bio/db/sam'
|
5
|
+
require "test/unit"
|
6
|
+
gem 'test-unit'
|
7
|
+
|
8
|
+
|
9
|
+
class TestBioDbSam < Test::Unit::TestCase
|
10
|
+
|
11
|
+
class << self
|
12
|
+
def shutdown
|
13
|
+
File.delete("test/samples/small/different_index.bam.bai")
|
14
|
+
File.delete("test/samples/small/dupes_rmdup.bam")
|
15
|
+
File.delete("test/samples/small/mates_fixed.bam")
|
16
|
+
File.delete("test/samples/small/reheader.bam")
|
17
|
+
File.delete("test/samples/small/test_chr.fasta.fai")
|
18
|
+
File.delete("test/samples/small/test_sorted.bam")
|
19
|
+
File.delete("test/samples/small/maps_merged.bam")
|
20
|
+
File.delete("test/samples/small/maps_cated.bam")
|
21
|
+
end
|
22
|
+
end
|
23
|
+
|
24
|
+
|
25
|
+
def setup
|
26
|
+
@test_folder = "test/samples/small"
|
27
|
+
@testTAMFile = @test_folder + "/test.tam"
|
28
|
+
@testBAMFile = @test_folder + "/testu.bam"
|
29
|
+
@testReference = @test_folder + "/test_chr.fasta"
|
30
|
+
@sam = Bio::DB::Sam.new(
|
31
|
+
:fasta => @testReference,
|
32
|
+
:bam => @testBAMFile
|
33
|
+
)
|
34
|
+
end
|
35
|
+
|
36
|
+
|
37
|
+
|
38
|
+
def test_new
|
39
|
+
assert_kind_of(Bio::DB::Sam, @sam)
|
40
|
+
end
|
41
|
+
|
42
|
+
def test_index
|
43
|
+
test_bai_file = @testBAMFile+".bai"
|
44
|
+
#test to see if the index file exists. If so, delete it
|
45
|
+
if File.exist?(test_bai_file) == true
|
46
|
+
puts "bam index exists....deleting..."
|
47
|
+
File.delete(test_bai_file)
|
48
|
+
end
|
49
|
+
#index the bam file
|
50
|
+
@sam.index()
|
51
|
+
|
52
|
+
#make sure the .bai file exists
|
53
|
+
assert_nothing_thrown do
|
54
|
+
File.open(test_bai_file, "r")
|
55
|
+
end
|
56
|
+
assert(File.size(test_bai_file) > 0, "From test_index: .bai file is empty")
|
57
|
+
#as above, but give the output a different name
|
58
|
+
test_bai_file = @test_folder+"/different_index.bam.bai"
|
59
|
+
@sam.index(:out_index=> test_bai_file)
|
60
|
+
assert_nothing_thrown do
|
61
|
+
File.open(test_bai_file, "r")
|
62
|
+
end
|
63
|
+
assert(File.size(test_bai_file) > 0, "From test_index: .bai file is empty")
|
64
|
+
end
|
65
|
+
|
66
|
+
def test_view
|
67
|
+
#how to get Bio::DB::Alignment objects ..
|
68
|
+
@sam.view() do |sam|
|
69
|
+
#test that all the objects are Bio::DB::Alignment objects and their reference is 'chr_1'
|
70
|
+
assert_equal(sam.class, Bio::DB::Alignment)
|
71
|
+
assert_equal(sam.rname, "chr_1")
|
72
|
+
end
|
73
|
+
end
|
74
|
+
|
75
|
+
def test_fetch
|
76
|
+
@sam.fetch("chr_1", 10,1000) do |sam|
|
77
|
+
#test that all the objects are Bio::DB::Alignment objects
|
78
|
+
assert_equal(sam.class, Bio::DB::Alignment)
|
79
|
+
assert_equal(sam.rname, "chr_1")
|
80
|
+
end
|
81
|
+
end
|
82
|
+
|
83
|
+
def test_fetch_with_function
|
84
|
+
#pass the assert to method
|
85
|
+
block = Proc.new {|a| assert_equal(a.class, Bio::DB::Alignment)}
|
86
|
+
@sam.fetch_with_function("chr_1", 10,1000, &block)
|
87
|
+
end
|
88
|
+
|
89
|
+
def test_chromosome_coverage
|
90
|
+
#the coverage should only be 1.0 or 2.0
|
91
|
+
cov = @sam.chromosome_coverage("chr_1", 33, 19)
|
92
|
+
cov.each do |pu|
|
93
|
+
assert_send([[1.0 , 2.0], :member?, pu])
|
94
|
+
end
|
95
|
+
end
|
96
|
+
|
97
|
+
def test_average_coverage
|
98
|
+
#there should be 10 positions with cov of 1.0 and 10 with cov of 2.0, so average of 1.5
|
99
|
+
test_bai_file = @testBAMFile+".bai"
|
100
|
+
if File.exist?(test_bai_file) == false
|
101
|
+
@sam.index()
|
102
|
+
end
|
103
|
+
avcov = @sam.average_coverage("chr_1", 33, 19)
|
104
|
+
assert_equal(avcov, 1.5)
|
105
|
+
File.delete(test_bai_file)
|
106
|
+
end
|
107
|
+
|
108
|
+
def test_faidx
|
109
|
+
@sam.faidx()
|
110
|
+
test_fai_file = @testReference+".fai"
|
111
|
+
#test that the .fai file exists
|
112
|
+
assert_nothing_thrown do
|
113
|
+
File.open(test_fai_file, "r")
|
114
|
+
end
|
115
|
+
#test that the file is not empty
|
116
|
+
assert(File.size(test_fai_file) > 0, "From test_faidx: .fai file is empty")
|
117
|
+
end
|
118
|
+
|
119
|
+
def test_index_stats
|
120
|
+
@sam.index_stats.each_pair do |seq, stat|
|
121
|
+
assert_send([['chr_1' , '*'], :member?, seq])
|
122
|
+
end
|
123
|
+
end
|
124
|
+
|
125
|
+
def test_fetch_reference
|
126
|
+
#this is the first 70 nucleotides of the test seqeunce
|
127
|
+
seq_expected = "CCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTA"
|
128
|
+
#fetch the first 70 nuclotides
|
129
|
+
seq_fetched = @sam.fetch_reference("chr_1", 1, 70, :as_bio => false)
|
130
|
+
#test they're the same
|
131
|
+
assert_equal(seq_fetched, seq_expected)
|
132
|
+
end
|
133
|
+
|
134
|
+
def test_sort
|
135
|
+
#sort the bam file
|
136
|
+
sortedsam = @test_folder + "/test_sorted.bam"
|
137
|
+
@sam.sort(:prefix=>@test_folder + "/test_sorted")
|
138
|
+
#create a new Bio::DB::Sam from the sorted bam
|
139
|
+
@sortsam = Bio::DB::Sam.new(
|
140
|
+
:fasta => @testReference,
|
141
|
+
:bam => sortedsam
|
142
|
+
)
|
143
|
+
pos = 0
|
144
|
+
#iterate over the sorted sam file and make sure that the it's sorted by checking the order of the start positions for each read.
|
145
|
+
@sortsam.view()do |sam|
|
146
|
+
assert(sam.pos > pos, "Not sorted by position")
|
147
|
+
pos = sam.pos
|
148
|
+
end
|
149
|
+
end
|
150
|
+
|
151
|
+
def test_reheader
|
152
|
+
sam_header = @test_folder + "/map_for_reheader.sam"
|
153
|
+
outfile = @test_folder + "/reheader.bam"
|
154
|
+
|
155
|
+
@sam.reheader(sam_header, :out=>outfile)
|
156
|
+
reheader_bam = Bio::DB::Sam.new(:fasta => @testReference, :bam => outfile)
|
157
|
+
#check that the reference is 'chr_2'
|
158
|
+
reheader_bam.view()do |sam|
|
159
|
+
assert_equal(sam.rname, "chr_2")
|
160
|
+
end
|
161
|
+
end
|
162
|
+
|
163
|
+
def test_calmd
|
164
|
+
no_md_sam = @test_folder + "/no_md.sam"
|
165
|
+
md = Bio::DB::Sam.new(:fasta => @testReference, :bam => no_md_sam)
|
166
|
+
block = Proc.new {|a| assert(a.tags.has_key?('MD'), "From test_calmd: couldn't find the MD tag")}
|
167
|
+
md.calmd(:S=>true, &block)
|
168
|
+
|
169
|
+
end
|
170
|
+
|
171
|
+
def test_mpileup
|
172
|
+
#create an mpileup
|
173
|
+
@sam.mpileup(:g => false) do |pileup|
|
174
|
+
#test that all the objects are Bio::DB::Pileup objects
|
175
|
+
assert_kind_of(Bio::DB::Pileup, pileup)
|
176
|
+
#test that the reference name is 'chr_1' for all objects
|
177
|
+
assert_equal(pileup.ref_name, 'chr_1')
|
178
|
+
end
|
179
|
+
#do the same for Vcf output
|
180
|
+
@sam.mpileup(:u => true) do |pileup|
|
181
|
+
assert_kind_of(Bio::DB::Vcf, pileup)
|
182
|
+
assert_equal(pileup.chrom, 'chr_1')
|
183
|
+
end
|
184
|
+
end
|
185
|
+
|
186
|
+
def test_mpileup_reg
|
187
|
+
#create an mpileup
|
188
|
+
reg = Bio::DB::Fasta::Region.new
|
189
|
+
reg.entry = "chr_1"
|
190
|
+
reg.start = 1
|
191
|
+
reg.end = 334
|
192
|
+
|
193
|
+
@sam.mpileup_cached(:r=>reg,:g => false, :min_cov => 1, :min_per =>0.2) do |pileup|
|
194
|
+
#test that all the objects are Bio::DB::Pileup objects
|
195
|
+
assert_kind_of(Bio::DB::Pileup, pileup)
|
196
|
+
#test that the reference name is 'chr_1' for all objects
|
197
|
+
#puts pileup
|
198
|
+
assert_equal(pileup.ref_name, 'chr_1')
|
199
|
+
end
|
200
|
+
|
201
|
+
region = @sam.cached_regions[reg.to_s]
|
202
|
+
#puts "cahced_region: #{region.inspect}"
|
203
|
+
puts "AVG COV: #{region.average_coverage}"
|
204
|
+
puts "Reference: #{region.reference}"
|
205
|
+
puts "Consensus: #{region.consensus}"
|
206
|
+
puts "called: #{region.called}"
|
207
|
+
#, :snps, :reference, :base_ratios, :consensus, :coverages
|
208
|
+
snps_tot = Bio::Sequence.snps_between(region.reference, region.consensus)
|
209
|
+
assert_equal(snps_tot, 5)
|
210
|
+
assert_equal(region.called, 213)
|
211
|
+
end
|
212
|
+
|
213
|
+
def test_mpileup_reg_05
|
214
|
+
#create an mpileup
|
215
|
+
reg = Bio::DB::Fasta::Region.new
|
216
|
+
reg.entry = "chr_1"
|
217
|
+
reg.start = 1
|
218
|
+
reg.end = 334
|
219
|
+
@sam.mpileup_cached(:r=>reg,:g => false, :min_cov => 1, :min_per =>0.4) do |pileup|
|
220
|
+
#test that all the objects are Bio::DB::Pileup objects
|
221
|
+
assert_kind_of(Bio::DB::Pileup, pileup)
|
222
|
+
#test that the reference name is 'chr_1' for all objects
|
223
|
+
#puts pileup
|
224
|
+
assert_equal(pileup.ref_name, 'chr_1')
|
225
|
+
|
226
|
+
end
|
227
|
+
|
228
|
+
region = @sam.cached_regions[reg.to_s]
|
229
|
+
|
230
|
+
#, :snps, :reference, :base_ratios, :consensus, :coverages
|
231
|
+
snps_tot = Bio::Sequence.snps_between(region.reference, region.consensus)
|
232
|
+
assert_equal(snps_tot, 1)
|
233
|
+
assert_equal(region.called, 213)
|
234
|
+
end
|
235
|
+
|
236
|
+
def test_depth
|
237
|
+
#the depth of coverage should be '1' at all given positions
|
238
|
+
@sam.depth(:r=>"chr_1:25-42") do |al|
|
239
|
+
assert_equal(al[2].to_i, 1)
|
240
|
+
end
|
241
|
+
end
|
242
|
+
|
243
|
+
def test_fixmate
|
244
|
+
mates_fixed_bam = @test_folder + "/mates_fixed.bam"
|
245
|
+
@sam.fix_mates(:out_bam=>mates_fixed_bam)
|
246
|
+
assert_nothing_thrown do
|
247
|
+
File.open(mates_fixed_bam, "r")
|
248
|
+
end
|
249
|
+
assert(File.size(mates_fixed_bam) > 0, "From test_fixmate: .bam file is empty")
|
250
|
+
end
|
251
|
+
|
252
|
+
def test_flagstats
|
253
|
+
#get the stats
|
254
|
+
stats = @sam.flag_stats()
|
255
|
+
#the number of reads mapped will be the first character on the first line.
|
256
|
+
no_reads_mapped = stats[0][0].to_i
|
257
|
+
#check that it's '9'
|
258
|
+
assert_equal(no_reads_mapped, 9)
|
259
|
+
end
|
260
|
+
|
261
|
+
def test_merge
|
262
|
+
bam1 = @test_folder + "/map_to_merge1.bam"
|
263
|
+
bam2 = @test_folder + "/map_to_merge2.bam"
|
264
|
+
bam_to_merge1 = Bio::DB::Sam.new(:fasta => @testReference, :bam => bam1)
|
265
|
+
bam_to_merge2 = Bio::DB::Sam.new(:fasta => @testReference, :bam => bam2)
|
266
|
+
bam_files = [bam_to_merge1, bam_to_merge2]
|
267
|
+
|
268
|
+
merged_bam_file = @test_folder + "/maps_merged.bam"
|
269
|
+
|
270
|
+
@sam.merge(:out=>merged_bam_file, :bams=>bam_files, :n=>true)
|
271
|
+
merged_bam = Bio::DB::Sam.new(:fasta => @testReference, :bam => merged_bam_file)
|
272
|
+
no_reads_mapped = 0;
|
273
|
+
|
274
|
+
merged_bam.view() do |al|
|
275
|
+
assert_kind_of(Bio::DB::Alignment, al)
|
276
|
+
no_reads_mapped+=1
|
277
|
+
end
|
278
|
+
assert_equal(no_reads_mapped, 10)
|
279
|
+
end
|
280
|
+
|
281
|
+
def test_cat
|
282
|
+
#same files used for merge, but we'll cat them instead
|
283
|
+
bam1 = @test_folder + "/map_to_merge1.bam"
|
284
|
+
bam2 = @test_folder + "/map_to_merge2.bam"
|
285
|
+
|
286
|
+
bam_files = [bam1, bam2]
|
287
|
+
|
288
|
+
cat_bam_file = @test_folder + "/maps_cated.bam"
|
289
|
+
|
290
|
+
@sam.merge(:out=>cat_bam_file, :bams=>bam_files)
|
291
|
+
cated_bam = Bio::DB::Sam.new(:fasta => @testReference, :bam => cat_bam_file)
|
292
|
+
|
293
|
+
no_reads_mapped = 0;
|
294
|
+
cated_bam.view() do |al|
|
295
|
+
assert_kind_of(Bio::DB::Alignment, al)
|
296
|
+
no_reads_mapped+=1
|
297
|
+
end
|
298
|
+
#there should be 10 reads in the cat'd maps
|
299
|
+
assert_equal(no_reads_mapped, 10)
|
300
|
+
end
|
301
|
+
|
302
|
+
def test_rmdup
|
303
|
+
#dupes contains 4 reads mapped once and one read mapped to the same place 268 times.
|
304
|
+
dupes = @test_folder + "/dupes.bam"
|
305
|
+
unduped = @test_folder + "/dupes_rmdup.bam"
|
306
|
+
bam_with_dupes = Bio::DB::Sam.new(:fasta => @testReference, :bam => dupes)
|
307
|
+
bam_with_dupes.remove_duplicates(:s=>true, :out=>unduped)
|
308
|
+
|
309
|
+
unduped_bam = Bio::DB::Sam.new(:fasta => @testReference, :bam => unduped)
|
310
|
+
#rmdup should remove 267 of the 268 reads mapping to the same place, so producing a bam file with 5 reads
|
311
|
+
readcount = 0
|
312
|
+
unduped_bam.view()do |sam|
|
313
|
+
readcount +=1
|
314
|
+
end
|
315
|
+
assert_equal(readcount, 5)
|
316
|
+
end
|
317
|
+
|
318
|
+
def test_targetcut
|
319
|
+
sorted_bam = @test_folder + "/sorted.bam"
|
320
|
+
cut = Bio::DB::Sam.new(:fasta => @testReference, :bam => sorted_bam)
|
321
|
+
assert_nothing_thrown do
|
322
|
+
cut.targetcut
|
323
|
+
end
|
324
|
+
end
|
325
|
+
|
326
|
+
def test_docs
|
327
|
+
#force an error (use 'samtool' instead of 'samtools')
|
328
|
+
output = Bio::DB::Sam.docs('samtool', 'tview')
|
329
|
+
assert_equal(output, "program must be 'samtools' or 'bcftools'")
|
330
|
+
end
|
331
|
+
end
|
data/test/test_vcf.rb
CHANGED
@@ -3,6 +3,7 @@ $: << File.expand_path('.')
|
|
3
3
|
require 'rubygems'
|
4
4
|
require 'bio/db/vcf'
|
5
5
|
require "test/unit"
|
6
|
+
gem 'test-unit'
|
6
7
|
|
7
8
|
|
8
9
|
class TestVcf < Test::Unit::TestCase
|
@@ -65,4 +66,14 @@ class TestVcf < Test::Unit::TestCase
|
|
65
66
|
"3"=>{"DP"=>"5", "GT"=>"1/1", "HQ"=>".,", "GQ"=>"43"}}, @vcf4.samples)
|
66
67
|
end
|
67
68
|
|
69
|
+
def test_int_or_raw
|
70
|
+
assert_nothing_raised do
|
71
|
+
@vcf1.int_or_raw(1)
|
72
|
+
end
|
73
|
+
assert_raise do
|
74
|
+
@vcf1.int_or_raw(A)
|
75
|
+
end
|
76
|
+
end
|
77
|
+
|
78
|
+
|
68
79
|
end
|
File without changes
|
File without changes
|
metadata
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: bio-samtools
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.
|
4
|
+
version: 2.0.3
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Ricardo Ramirez-Gonzalez
|
@@ -10,22 +10,22 @@ authors:
|
|
10
10
|
autorequire:
|
11
11
|
bindir: bin
|
12
12
|
cert_chain: []
|
13
|
-
date: 2014-
|
13
|
+
date: 2014-04-18 00:00:00.000000000 Z
|
14
14
|
dependencies:
|
15
15
|
- !ruby/object:Gem::Dependency
|
16
|
-
name:
|
16
|
+
name: bio-svgenes
|
17
17
|
requirement: !ruby/object:Gem::Requirement
|
18
18
|
requirements:
|
19
19
|
- - ">="
|
20
20
|
- !ruby/object:Gem::Version
|
21
|
-
version:
|
21
|
+
version: 0.4.1
|
22
22
|
type: :runtime
|
23
23
|
prerelease: false
|
24
24
|
version_requirements: !ruby/object:Gem::Requirement
|
25
25
|
requirements:
|
26
26
|
- - ">="
|
27
27
|
- !ruby/object:Gem::Version
|
28
|
-
version:
|
28
|
+
version: 0.4.1
|
29
29
|
- !ruby/object:Gem::Dependency
|
30
30
|
name: bio
|
31
31
|
requirement: !ruby/object:Gem::Requirement
|
@@ -40,20 +40,6 @@ dependencies:
|
|
40
40
|
- - ">="
|
41
41
|
- !ruby/object:Gem::Version
|
42
42
|
version: 1.4.2
|
43
|
-
- !ruby/object:Gem::Dependency
|
44
|
-
name: systemu
|
45
|
-
requirement: !ruby/object:Gem::Requirement
|
46
|
-
requirements:
|
47
|
-
- - ">="
|
48
|
-
- !ruby/object:Gem::Version
|
49
|
-
version: 2.5.2
|
50
|
-
type: :runtime
|
51
|
-
prerelease: false
|
52
|
-
version_requirements: !ruby/object:Gem::Requirement
|
53
|
-
requirements:
|
54
|
-
- - ">="
|
55
|
-
- !ruby/object:Gem::Version
|
56
|
-
version: 2.5.2
|
57
43
|
- !ruby/object:Gem::Dependency
|
58
44
|
name: shoulda
|
59
45
|
requirement: !ruby/object:Gem::Requirement
|
@@ -166,6 +152,20 @@ dependencies:
|
|
166
152
|
- - ">="
|
167
153
|
- !ruby/object:Gem::Version
|
168
154
|
version: '0'
|
155
|
+
- !ruby/object:Gem::Dependency
|
156
|
+
name: test-unit
|
157
|
+
requirement: !ruby/object:Gem::Requirement
|
158
|
+
requirements:
|
159
|
+
- - ">="
|
160
|
+
- !ruby/object:Gem::Version
|
161
|
+
version: '0'
|
162
|
+
type: :development
|
163
|
+
prerelease: false
|
164
|
+
version_requirements: !ruby/object:Gem::Requirement
|
165
|
+
requirements:
|
166
|
+
- - ">="
|
167
|
+
- !ruby/object:Gem::Version
|
168
|
+
version: '0'
|
169
169
|
description: "Binder of samtools for ruby, on the top of FFI. \n\n This project was
|
170
170
|
born from the need to add support of BAM files to \n the gee_fu genome browser
|
171
171
|
(http://github.com/danmaclean/gee_fu)."
|
@@ -192,80 +192,19 @@ files:
|
|
192
192
|
- doc/Bio/DB/SAM.html
|
193
193
|
- doc/Bio/DB/SAM/Library.html
|
194
194
|
- doc/Bio/DB/SAM/Tools.html
|
195
|
-
- doc/Bio/DB/SAM/Tools/Bam1CoreT.html
|
196
|
-
- doc/Bio/DB/SAM/Tools/Bam1T.html
|
197
|
-
- doc/Bio/DB/SAM/Tools/BamHeaderT.html
|
198
|
-
- doc/Bio/DB/SAM/Tools/BamPileup1T.html
|
199
|
-
- doc/Bio/DB/SAM/Tools/SamfileT.html
|
200
|
-
- doc/Bio/DB/SAM/Tools/SamfileTX.html
|
201
|
-
- doc/Bio/DB/SAMException.html
|
202
195
|
- doc/Bio/DB/Tag.html
|
203
196
|
- doc/Bio/DB/Vcf.html
|
204
197
|
- doc/LICENSE_txt.html
|
205
|
-
- doc/LibC.html
|
206
|
-
- doc/Pileup.html
|
207
|
-
- doc/Vcf.html
|
208
|
-
- doc/basic_styles.css
|
209
|
-
- doc/classes/Bio.html
|
210
|
-
- doc/classes/Bio/DB.html
|
211
|
-
- doc/classes/Bio/DB/Alignment.html
|
212
|
-
- doc/classes/Bio/DB/Alignment.src/M000012.html
|
213
|
-
- doc/classes/Bio/DB/Alignment.src/M000013.html
|
214
|
-
- doc/classes/Bio/DB/Alignment.src/M000014.html
|
215
|
-
- doc/classes/Bio/DB/Alignment.src/M000015.html
|
216
|
-
- doc/classes/Bio/DB/SAM.html
|
217
|
-
- doc/classes/Bio/DB/SAM/Library.html
|
218
|
-
- doc/classes/Bio/DB/SAM/Library.src/M000006.html
|
219
|
-
- doc/classes/Bio/DB/SAM/Tools.html
|
220
|
-
- doc/classes/Bio/DB/SAM/Tools.src/M000007.html
|
221
|
-
- doc/classes/Bio/DB/SAM/Tools/Bam1CoreT.html
|
222
|
-
- doc/classes/Bio/DB/SAM/Tools/Bam1T.html
|
223
|
-
- doc/classes/Bio/DB/SAM/Tools/Bam1T.src/M000010.html
|
224
|
-
- doc/classes/Bio/DB/SAM/Tools/BamHeaderT.html
|
225
|
-
- doc/classes/Bio/DB/SAM/Tools/BamHeaderT.src/M000008.html
|
226
|
-
- doc/classes/Bio/DB/SAM/Tools/BamHeaderT.src/M000009.html
|
227
|
-
- doc/classes/Bio/DB/SAM/Tools/BamPileup1T.html
|
228
|
-
- doc/classes/Bio/DB/SAM/Tools/SamfileT.html
|
229
|
-
- doc/classes/Bio/DB/SAM/Tools/SamfileTX.html
|
230
|
-
- doc/classes/Bio/DB/SAMException.html
|
231
|
-
- doc/classes/Bio/DB/SAMException.src/M000016.html
|
232
|
-
- doc/classes/Bio/DB/Sam.src/M000017.html
|
233
|
-
- doc/classes/Bio/DB/Sam.src/M000018.html
|
234
|
-
- doc/classes/Bio/DB/Sam.src/M000019.html
|
235
|
-
- doc/classes/Bio/DB/Sam.src/M000020.html
|
236
|
-
- doc/classes/Bio/DB/Sam.src/M000021.html
|
237
|
-
- doc/classes/Bio/DB/Sam.src/M000022.html
|
238
|
-
- doc/classes/Bio/DB/Sam.src/M000023.html
|
239
|
-
- doc/classes/Bio/DB/Sam.src/M000024.html
|
240
|
-
- doc/classes/Bio/DB/Sam.src/M000025.html
|
241
|
-
- doc/classes/Bio/DB/Sam.src/M000026.html
|
242
|
-
- doc/classes/Bio/DB/Sam.src/M000027.html
|
243
|
-
- doc/classes/Bio/DB/Sam.src/M000028.html
|
244
|
-
- doc/classes/Bio/DB/Sam.src/M000029.html
|
245
|
-
- doc/classes/Bio/DB/Sam.src/M000030.html
|
246
|
-
- doc/classes/Bio/DB/Sam.src/M000031.html
|
247
|
-
- doc/classes/Bio/DB/Sam.src/M000032.html
|
248
|
-
- doc/classes/Bio/DB/Tag.html
|
249
|
-
- doc/classes/Bio/DB/Tag.src/M000011.html
|
250
|
-
- doc/classes/LibC.html
|
251
|
-
- doc/classes/Pileup.html
|
252
|
-
- doc/classes/Pileup.src/M000001.html
|
253
|
-
- doc/classes/Pileup.src/M000002.html
|
254
|
-
- doc/classes/Pileup.src/M000003.html
|
255
|
-
- doc/classes/Pileup.src/M000004.html
|
256
|
-
- doc/classes/Pileup.src/M000005.html
|
257
198
|
- doc/created.rid
|
258
|
-
- doc/
|
259
|
-
- doc/
|
260
|
-
- doc/
|
261
|
-
- doc/
|
262
|
-
- doc/
|
263
|
-
- doc/
|
264
|
-
- doc/
|
265
|
-
- doc/fr_class_index.html
|
266
|
-
- doc/fr_file_index.html
|
267
|
-
- doc/fr_method_index.html
|
199
|
+
- doc/fonts.css
|
200
|
+
- doc/fonts/Lato-Light.ttf
|
201
|
+
- doc/fonts/Lato-LightItalic.ttf
|
202
|
+
- doc/fonts/Lato-Regular.ttf
|
203
|
+
- doc/fonts/Lato-RegularItalic.ttf
|
204
|
+
- doc/fonts/SourceCodePro-Bold.ttf
|
205
|
+
- doc/fonts/SourceCodePro-Regular.ttf
|
268
206
|
- doc/images/add.png
|
207
|
+
- doc/images/arrow_up.png
|
269
208
|
- doc/images/brick.png
|
270
209
|
- doc/images/brick_link.png
|
271
210
|
- doc/images/bug.png
|
@@ -296,64 +235,67 @@ files:
|
|
296
235
|
- doc/js/search.js
|
297
236
|
- doc/js/search_index.js
|
298
237
|
- doc/js/searcher.js
|
299
|
-
- doc/lib/bio-samtools_rb.html
|
300
|
-
- doc/lib/bio/db/pileup_rb.html
|
301
|
-
- doc/lib/bio/db/sam/bam_rb.html
|
302
|
-
- doc/lib/bio/db/sam/faidx_rb.html
|
303
|
-
- doc/lib/bio/db/sam/library_rb.html
|
304
|
-
- doc/lib/bio/db/sam/pileup_rb.html
|
305
|
-
- doc/lib/bio/db/sam/sam_rb.html
|
306
|
-
- doc/lib/bio/db/sam/vcf_rb.html
|
307
|
-
- doc/lib/bio/db/sam_rb.html
|
308
|
-
- doc/lib/bio/db/vcf_rb.html
|
309
|
-
- doc/rdoc-style.css
|
310
238
|
- doc/rdoc.css
|
311
239
|
- doc/table_of_contents.html
|
312
|
-
- doc/tutorial.html
|
313
|
-
- doc/tutorial.pdf
|
314
240
|
- ext/Makefile-bioruby.patch
|
315
241
|
- ext/Makefile-suse.patch
|
242
|
+
- ext/Rakefile
|
316
243
|
- ext/mkrf_conf.rb
|
317
244
|
- lib/bio-samtools.rb
|
318
245
|
- lib/bio/.DS_Store
|
246
|
+
- lib/bio/BIOExtensions.rb
|
247
|
+
- lib/bio/db/alignment.rb
|
248
|
+
- lib/bio/db/fastadb.rb
|
319
249
|
- lib/bio/db/pileup.rb
|
320
250
|
- lib/bio/db/sam.rb
|
321
|
-
- lib/bio/db/sam/bam.rb
|
322
251
|
- lib/bio/db/sam/external/COPYING
|
323
252
|
- lib/bio/db/sam/external/VERSION
|
324
|
-
- lib/bio/db/sam/
|
253
|
+
- lib/bio/db/sam/faidx_old.rb
|
325
254
|
- lib/bio/db/sam/library.rb
|
326
|
-
- lib/bio/db/sam/sam.rb
|
327
255
|
- lib/bio/db/vcf.rb
|
256
|
+
- test/.gitignore
|
328
257
|
- test/helper.rb
|
329
|
-
- test/
|
330
|
-
- test/samples/
|
331
|
-
- test/samples/
|
332
|
-
- test/samples/pipe_char/test_chr.fasta
|
333
|
-
- test/samples/pipe_char/test_chr.fasta.fai
|
258
|
+
- test/old_test_basic.rb
|
259
|
+
- test/samples/small/dupes.bam
|
260
|
+
- test/samples/small/dupes.sam
|
334
261
|
- test/samples/small/ids2.txt
|
262
|
+
- test/samples/small/map_for_reheader.sam
|
263
|
+
- test/samples/small/map_to_merge1.bam
|
264
|
+
- test/samples/small/map_to_merge1.bam.bai
|
265
|
+
- test/samples/small/map_to_merge1.sam
|
266
|
+
- test/samples/small/map_to_merge2.bam
|
267
|
+
- test/samples/small/map_to_merge2.bam.bai
|
268
|
+
- test/samples/small/map_to_merge2.sam
|
269
|
+
- test/samples/small/no_md.sam
|
335
270
|
- test/samples/small/sorted.bam
|
336
271
|
- test/samples/small/sorted.bam.bai
|
337
|
-
- test/samples/small/test
|
338
|
-
- test/samples/small/test.bam
|
339
|
-
- test/samples/small/test.fa
|
340
|
-
- test/samples/small/test.fai
|
341
272
|
- test/samples/small/test.sai
|
342
273
|
- test/samples/small/test.tam
|
343
274
|
- test/samples/small/test_chr.fasta
|
275
|
+
- test/samples/small/test_chr.fasta.1.bt2
|
276
|
+
- test/samples/small/test_chr.fasta.2.bt2
|
277
|
+
- test/samples/small/test_chr.fasta.3.bt2
|
278
|
+
- test/samples/small/test_chr.fasta.4.bt2
|
344
279
|
- test/samples/small/test_chr.fasta.amb
|
345
280
|
- test/samples/small/test_chr.fasta.ann
|
346
281
|
- test/samples/small/test_chr.fasta.bwt
|
347
282
|
- test/samples/small/test_chr.fasta.pac
|
348
283
|
- test/samples/small/test_chr.fasta.rbwt
|
284
|
+
- test/samples/small/test_chr.fasta.rev.1.bt2
|
285
|
+
- test/samples/small/test_chr.fasta.rev.2.bt2
|
349
286
|
- test/samples/small/test_chr.fasta.rpac
|
350
287
|
- test/samples/small/test_chr.fasta.rsa
|
351
288
|
- test/samples/small/test_chr.fasta.sa
|
289
|
+
- test/samples/small/test_cov.svg
|
352
290
|
- test/samples/small/testu.bam
|
353
291
|
- test/samples/small/testu.bam.bai
|
354
|
-
- test/
|
292
|
+
- test/svg
|
355
293
|
- test/test_bio-samtools.rb
|
294
|
+
- test/test_pileup.rb
|
295
|
+
- test/test_sam.rb
|
356
296
|
- test/test_vcf.rb
|
297
|
+
- tutorial/tutorial.html
|
298
|
+
- tutorial/tutorial.pdf
|
357
299
|
homepage: http://github.com/helios/bioruby-samtools
|
358
300
|
licenses:
|
359
301
|
- MIT
|