rbbt-util 5.43.0 → 5.44.1

data/test/rbbt/tsv/test_parser.rb

@@ -1,101 +0,0 @@
-require File.expand_path(File.join(File.dirname(__FILE__), '../..', 'test_helper.rb'))
-require 'rbbt/tsv'
-
-class TestTSVParser < Test::Unit::TestCase
-  def test_flat
-    TmpFile.with_file do |tmpdir|
-      file = File.join(tmpdir, 'file')
-      str =<<-EOF
-#: :type=:flat#:sep=' '
-#Key Value
-a b c d e
-A B C D E
-      EOF
-      Open.write(file, str)
-      Open.open(file) do |io|
-        parser = TSV::Parser.new(io, :merge => true, :zipped => true)
-        parser.traverse do |k,v|
-          assert v.length > 1
-        end
-      end
-    end
-  end
-
-  def test_traverse
-
-    content =<<-EOF
-row1    a|aa|aaa    b    Id1|Id2
-    EOF
-
-    TmpFile.with_file(content) do |filename|
-      TSV::Parser.traverse(Open.open(filename), :sep => /\s+/, :fields => [1], :type => :double) do |k,v|
-        assert_equal ["row1"], k
-        assert_equal [%w(a aa aaa)], v
-      end
-
-      TSV::Parser.traverse(Open.open(filename), :sep => /\s+/, :fields => [1], :type => :list) do |k,v|
-        assert_equal "row1", k
-        assert_equal ["a|aa|aaa"], v
-      end
-
-      TSV::Parser.traverse(Open.open(filename), :sep => /\s+/, :fields => [1], :type => :single) do |k,v|
-        assert_equal "row1", k
-        assert_equal "a|aa|aaa", v
-      end
-
-      TSV::Parser.traverse(Open.open(filename), :sep => /\s+/, :fields => [1], :type => :flat) do |k,v|
-        assert_equal ["row1"], k
-        assert_equal %w(a aa aaa), v
-      end
-    end
-  end
-
-  def test_fields
-    TmpFile.with_file do |tmpdir|
-      file = File.join(tmpdir, 'file')
-      str =<<-EOF
-#: :type=:flat#:sep=' '
-#Key Value
-a b c d e
-A B C D E
-      EOF
-      Open.write(file, str)
-      Open.open(file) do |io|
-        parser = TSV::Parser.new(io, :merge => true, :zipped => true, :type => :double)
-        parser.traverse do |k,v,f|
-          assert_equal f, %w(Value)
-        end
-      end
-    end
-  end
-
-  def test_merge
-    content =<<-EOF
-#: :type=:double
-#PMID:Sentence number:TF:TG	Transcription Factor (Associated Gene Name)	Target Gene (Associated Gene Name)	Sign	Negation	PMID
-24265317:3:NR1H3:FASN	NR1H3	FASN			24265317
-17522048:0:NR1H3:FASN	NR1H3	FASN	+		17522048
-19903962:0:NR1H3:FASN	NR1H3	FASN			19903962
-19903962:7:NR1H3:FASN	NR1H3	FASN			19903962
-22183856:4:NR1H3:FASN	NR1H3	FASN			22183856
-22641099:4:NR1H3:FASN	NR1H3	FASN	+		22641099
-23499676:8:NR1H3:FASN	NR1H3	FASN	+		23499676
-11790787:5:NR1H3:FASN	NR1H3	FASN			11790787
-11790787:7:NR1H3:FASN	NR1H3	FASN	+		11790787
-11790787:9:NR1H3:FASN	NR1H3	FASN	+		11790787
-11790787:11:NR1H3:FASN	NR1H3	FASN			11790787
-17522048:1:NR1H3:FASN	NR1H3	FASN	+		17522048
-17522048:3:NR1H3:FASN	NR1H3	FASN			17522048
-22160584:1:NR1H3:FASN	NR1H3	FASN			22160584
-22160584:5:NR1H3:FASN	NR1H3	FASN	+		22160584
-22160584:8:NR1H3:FASN	NR1H3	FASN	+		22160584
-    EOF
-
-    TmpFile.with_file(content) do |filename|
-      tsv = TSV.open(filename, :key_field => "Transcription Factor (Associated Gene Name)", :fields => ["Target Gene (Associated Gene Name)", "Sign", "PMID"], :merge => true, :type => :double)
-      assert_equal 16, tsv["NR1H3"]["Sign"].length
-    end
-  end
-
-end
-

data/test/rbbt/tsv/test_stream.rb

@@ -1,253 +0,0 @@
-require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helper.rb')
-require 'rbbt/tsv'
-require 'rbbt/tsv/stream'
-require 'rbbt'
-
-class TestStream < Test::Unit::TestCase
-
-  def test_collapse_stream
-    text=<<-EOF
-#: :sep=" "
-#Row LabelA LabelB LabelC
-row1 A B C
-row1 a b c
-row2 AA BB CC
-row2 aa bb cc
-    EOF
-
-    s = StringIO.new text
-    tsv = TSV.open TSV.collapse_stream(s)
-    assert_equal ["A", "a"], tsv["row1"][0]
-    assert_equal ["BB", "bb"], tsv["row2"][1]
-  end
-
-  def test_paste_stream
-    text1=<<-EOF
-#: :sep=" "
-#Row LabelA LabelB LabelC
-row1 A B C
-row2 AA BB CC
-row3 AAA BBB CCC
-    EOF
-
-    text2=<<-EOF
-#: :sep=" "
-#Row Labela Labelb 
-row1 a b
-row2 aa bb
-row3 aaa bbb
-    EOF
-
-    text3=<<-EOF
-#: :sep=" "
-#Row LabelC
-row1 c
-row2 cc
-row3 ccc
-    EOF
-
-    s1 = StringIO.new text1
-    s2 = StringIO.new text2
-    s3 = StringIO.new text3
-    tsv = TSV.open TSV.paste_streams([s1,s2,s3], :sep => " ", :type => :list)
-    assert_equal ["A", "B", "C", "a", "b", "c"], tsv["row1"]
-    assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
-    assert_equal ["AAA", "BBB", "CCC", "aaa", "bbb", "ccc"], tsv["row3"]
-  end
-
-  def test_paste_stream_sort
-    text1=<<-EOF
-#: :sep=" "
-#Row LabelA LabelB LabelC
-row2 AA BB CC
-row1 A B C
-row3 AAA BBB CCC
-    EOF
-
-    text2=<<-EOF
-#: :sep=" "
-#Row Labela Labelb
-row1 a b
-row3 aaa bbb
-row2 aa bb
-    EOF
-
-    text3=<<-EOF
-#: :sep=" "
-#Row Labelc
-row3 ccc
-row1 c
-row2 cc
-    EOF
-
-    s1 = StringIO.new text1
-    s2 = StringIO.new text2
-    s3 = StringIO.new text3
-    tsv = TSV.open TSV.paste_streams([s1,s2,s3], :sep => " ", :type => :list, :sort => true)
-    assert_equal "Row", tsv.key_field
-    assert_equal %w(LabelA LabelB LabelC Labela Labelb Labelc), tsv.fields
-    assert_equal ["A", "B", "C", "a", "b", "c"], tsv["row1"]
-    assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
-    assert_equal ["AAA", "BBB", "CCC", "aaa", "bbb", "ccc"], tsv["row3"]
-  end
-
-  def test_paste_stream_missing_2
-    text1=<<-EOF
-#: :sep=" "
-#Row LabelA LabelB LabelC
-row2 AA BB CC
-row1 A B C
-    EOF
-
-    text2=<<-EOF
-#: :sep=" "
-#Row Labela Labelb
-row2 aa bb
-    EOF
-
-    text3=<<-EOF
-#: :sep=" "
-#Row Labelc
-row3 ccc
-row2 cc
-    EOF
-
-    s1 = StringIO.new text1
-    s2 = StringIO.new text2
-    s3 = StringIO.new text3
-    tsv = TSV.open TSV.paste_streams([s1,s2,s3], :sep => " ", :type => :list, :sort => true)
-    assert_equal "Row", tsv.key_field
-    assert_equal %w(LabelA LabelB LabelC Labela Labelb Labelc), tsv.fields
-    assert_equal ["A", "B", "C", "", "", ""], tsv["row1"]
-    assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
-    assert_equal ["", "", "", "", "", "ccc"], tsv["row3"]
-  end
-
-  def test_paste_stream_missing
-    text1=<<-EOF
-#: :sep=" "
-#Row LabelA LabelB LabelC
-row2 AA BB CC
-row1 A B C
-    EOF
-
-    text2=<<-EOF
-#: :sep=" "
-#Row Labela Labelb
-row2 aa bb
-    EOF
-
-    text3=<<-EOF
-#: :sep=" "
-#Row Labelc
-row3 ccc
-row2 cc
-    EOF
-
-    s1 = StringIO.new text1
-    s2 = StringIO.new text2
-    s3 = StringIO.new text3
-    tsv = TSV.open TSV.paste_streams([s1,s2,s3], :sep => " ", :type => :list, :sort => true)
-    assert_equal "Row", tsv.key_field
-    assert_equal %w(LabelA LabelB LabelC Labela Labelb Labelc), tsv.fields
-    assert_equal ["A", "B", "C", "", "", ""], tsv["row1"]
-    assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
-    assert_equal ["", "", "", "", "", "ccc"], tsv["row3"]
-  end
-
-  def test_paste_stream_missing_3
-    text1=<<-EOF
-#: :sep=" "
-#Row LabelA LabelB LabelC
-row2 AA BB CC
-row1 A B C
-    EOF
-
-    text2=<<-EOF
-#: :sep=" "
-#Row Labelc
-    EOF
-
-    s1 = StringIO.new text1
-    s2 = StringIO.new text2
-    tsv = TSV.open TSV.paste_streams([s1,s2], :sep => " ", :type => :list, :sort => true)
-    assert_equal "Row", tsv.key_field
-    assert_equal %w(LabelA LabelB LabelC Labelc), tsv.fields
-    assert_equal ["A", "B", "C", ""], tsv["row1"]
-    assert_equal ["AA", "BB", "CC", ""], tsv["row2"]
-  end
-
-  def test_paste_stream_same_field
-    text1=<<-EOF
-#: :sep=" "
-#Row LabelA
-row1 A
-row2 AA
-    EOF
-
-    text2=<<-EOF
-#: :sep=" "
-#Row LabelA
-row2 AAA
-    EOF
-
-    s1 = StringIO.new text1
-    s2 = StringIO.new text2
-    tsv = TSV.open TSV.paste_streams([s1,s2], :sep => " ", :type => :double, :sort => false, :same_fields => true)
-    assert_equal "Row", tsv.key_field
-    assert_equal ["AA", "AAA"], tsv["row2"][0]
-  end
-
-  def test_paste_stream_nohead
-    text1=<<-EOF
-row1\tA
-row2\tAA
-    EOF
-
-    text2=<<-EOF
-row2\tAAA
-    EOF
-
-    s1 = StringIO.new text1
-    s2 = StringIO.new text2
-    tsv = TSV.open TSV.paste_streams([s1,s2], :type => :double, :sort => false, :same_fields => true)
-    assert_equal ["AA", "AAA"], tsv["row2"][0]
-  end
-
-  def test_flat2double
-    text1=<<-EOF
-#: :sep= #:type=:flat
-#Row LabelA
-row1 A AA AAA
-row2 a aa aaa
-    EOF
-
-    text2=<<-EOF
-#: :sep= #:type=:double
-#Row LabelA
-row1 A|AA|AAA
-row2 a|aa|aaa
-    EOF
-    s1 = StringIO.new text1
-    s2 = TSV.stream_flat2double(s1, :sep => " ")
-
-    assert_equal text2, s2.stream.read
-  end
-
-  def test_reorder_stream
-    text=<<-EOF
-#: :sep=" "
-#Row LabelA LabelB LabelC
-row1 A B C
-row1 a b c
-row2 AA BB CC
-row2 aa bb cc
-row4  BBB CC
-    EOF
-
-    s = StringIO.new text
-    dumper = TSV.reorder_stream_tsv(s, "LabelC", %w(Row LabelA))
-    tsv = TSV.open TSV.collapse_stream(dumper.stream)
-    assert_equal %w(row2 row4), tsv["CC"]["Row"]
-  end
-end

data/test/rbbt/tsv/test_util.rb

@@ -1,52 +0,0 @@
-require File.expand_path(File.dirname(__FILE__) + '/../../test_helper')
-require 'rbbt/tsv'
-require 'rbbt/tsv/util'
-require 'rbbt/util/tmpfile'
-require 'test/unit'
-
-class TestTSVUtil < Test::Unit::TestCase
-
-  def test_field_counts
-   content =<<-EOF
-#Id    ValueA    ValueB    OtherID
-row1    a|aa|aaa    b    Id1|Id2
-row2    A    B    Id3
-    EOF
-
-    TmpFile.with_file(content) do |filename|
-      tsv = TSV.open(filename, :sep => /\s+/)
-
-      assert_equal 2, TSV.field_match_counts(tsv, ["a","A","a","b","Id3"])["ValueA"]
-      assert_equal nil, TSV.field_match_counts(tsv, ["ValueA"])["ValueA"]
-    end
-  end
-
-  def test_replicates
-   content =<<-EOF
-#Id    ValueA    ValueB    OtherID
-row1    a|aa|aaa    b|bb|bbb    Id1|Id2|Id3
-row2    A    B    Id3
-    EOF
-
-    TmpFile.with_file(content) do |filename|
-      tsv = TSV.open(filename, :sep => /\s+/)
-
-      assert_equal 4, tsv.unzip_replicates.length
-      assert_equal %w(aa bb Id2), tsv.unzip_replicates["row1(1)"]
-    end
-  end
-
-  def test_merge
-   content =<<-EOF
-#Id    ValueA    ValueB    OtherID
-row1    a|aa|aaa    b|bb|bbb    Id1|Id2|Id3
-row2    A    B    Id3
-    EOF
-
-    TmpFile.with_file(content) do |filename|
-      tsv = TSV.open(filename, :sep => /\s+/)
-      tsv = tsv.merge({"row3" => [["A3"], ["B3"], ["Id4"]]})
-      assert TSV === tsv
-    end
-  end
-end

data/test/rbbt/util/R/test_eval.rb

@@ -1,43 +0,0 @@
-require File.join(File.expand_path(File.dirname(__FILE__)), '../../..', 'test_helper.rb')
-require 'rbbt/util/R/eval'
-
-class TestREval < Test::Unit::TestCase
-  def test_get
-    Misc.benchmark(1000) do
-      a = R.eval <<-EOF
-    p = 12
-    a = p * 2
-    c(a)
-      EOF
-      assert_equal 12 * 2, a
-    end
-  end
-
-  def test_TSV_fork
-    tsv = TSV.setup({"1" => "1"},:type => :single)
-      a = tsv.R <<-EOF, :R_method => :eval
-    p = 12
-    a = p * 2
-    c(a)
-      EOF
-
-      pid = Process.fork do
-        a = tsv.R <<-EOF, :R_method => :eval
-    p = 12
-    a = p * 2
-    c(a)
-        EOF
-      end
-      s = Process.waitpid2 pid
-  end
-
-  def test_TSV
-    tsv = TSV.setup({"1" => "1"},:type => :single)
-      a = tsv.R <<-EOF, :R_method => :eval
-    p = 12
-    a = p * 2
-    c(a)
-      EOF
-  end
-end
-

data/test/rbbt/util/R/test_model.rb

@@ -1,128 +0,0 @@
-require File.join(File.expand_path(File.dirname(__FILE__)), '../../..', 'test_helper.rb')
-require 'rbbt/util/R'
-require 'rbbt/util/R/model'
-
-class TestRModel < Test::Unit::TestCase
-
-  def test_fit
-    data = TSV.setup({}, :key_field => "Dose", :fields => ["Response"], :type => :single)
-    10.times do 
-      x = rand(10)
-      y = 10 + 3 * x + rand * 4
-      data[x] = y
-    end
-
-    model = R::Model.new "Test fit 2", "Response ~ Dose"
-
-    model.fit(data, method='drm', :fct => "LL.3()") unless model.exists?
-
-    x = 5
-    y = 10 + 3 * x 
-    input = TSV.setup({"new 1" => [x]}, :key_field => "Code", :fields => ["Dose"], :type => :single)
-    pred = model.predict(input)["new 1"]["Prediction"].to_f
-    assert pred > y and pred < y + 4
-  end
-
-  def test_add_fit
-    tsv = TSV.open datafile_test('dose_response'), :type => :list
-    tsv = tsv.slice(["Dose", "Response"])
-
-    result = tsv.R <<-EOF, :R_method => :shell
-library(drc, quietly=T)
-library(txtplot)
-data = rbbt.model.add_fit(data, Response ~ Dose, method=drm, classes='numeric', fct=LL.4(),na.action=na.omit)
-txtplot(data$Dose, data$Response)
-txtplot(data$Dose, data$Prediction)
-    EOF
-    assert result.fields.include? "Prediction"
-  end
-
-  def test_add_inpute
-    tsv = TSV.open datafile_test('dose_response'), :type => :list
-
-    result = tsv.R <<-EOF, :R_method => :eval
-library(drc, quietly=T)
-data = rbbt.model.inpute(data, CI ~ Dose, method=drm, classes='numeric', fct=LL.4(), na.action=na.exclude)
-    EOF
-
-    assert_equal result.size, result.column("CI").values.flatten.reject{|p| p.nil? or p.empty? or p == "NA"}.length
-  end
-
-  def __test_ab_surv_corr
-    require 'rbbt/workflow'
-    Workflow.require_workflow "Miller"
-
-    require 'rbbt/util/R/model'
-    require 'rbbt/util/R/svg'
-
-    antibody = "14-3-3-Zeta(C)_GBL9006927"
-
-    rppa = Miller.RPPA.data.tsv
-    rppa.attach Miller.RPPA.labels
-    viability = Miller.Viability.data.tsv
-
-    viability.add_field "Perturbation" do |compound,values|
-      values["Dose"].collect do |dose|
-        compound.split("-").flatten.zip(dose.split("-")).collect{|p| p * "="} * "-"
-      end
-    end
-
-    viability = viability.reorder "Perturbation", nil, :zipped => true
-    compounds = viability.column("Compound").flatten.uniq
-
-    rppa = rppa.select("Compound"){|c| ! c.include? "-" and compounds.include? c}
-    rppa = rppa.slice([antibody,"Compound", "Dose"])
-    rppa.rename_field antibody, "RPPA"
-
-    model = R::Model.new "viability", "Effect ~ Dose", nil, "Compound" => :factor
-
-    model.fit(viability.select("Compound"){|c| ! c.include? "-"}, 'lm')
-
-    rppa = model.predict(rppa, "Prediction")
-
-    plot_script = "plot<-ggplot(data=data) + geom_point(aes(x=RPPA, y=Prediction, color=Compound));"
-
-    R::SVG.ggplotSVG rppa, plot_script, 7, 7, :R_method => :eval 
-  end
-
-  def test_fit_fast
-    data = TSV.setup({}, :key_field => "Dose", :fields => ["Response"], :type => :single)
-    10.times do 
-      x = rand(10)
-      y = 10 + 3 * x + rand * 4
-      data[x] = y
-    end
-
-    model = R::Model.new "Test fit 2", "Response ~ Dose", data, :fit => 'lm'
-
-    x = 5
-    y = 10 + 3 * x 
-
-    pred = model.predict x
-    assert pred > y and pred < y + 4
-
-    pred = model.predict [x, 2*x, 3*x]
-    assert pred.first > y and pred.first  < y + 4
-
-
-    pred = model.predict "Dose" => x
-    assert pred > y and pred < y + 4
-  end
-
-  def test_interval
-    data = TSV.setup({}, :key_field => "Dose", :fields => ["Response"], :type => :single)
-    10.times do 
-      x = rand(10)
-      y = 10 + 3 * x + rand * 4
-      data[x] = y
-    end
-
-    model = R::Model.new "Test fit 2", "Response ~ Dose", data, :fit => 'lm'
-
-    x = 5
-    y = 10 + 3 * x 
-
-    pred = model.predict x
-    assert pred > y and pred < y + 4
-  end
-end

data/test/rbbt/util/R/test_plot.rb

@@ -1,38 +0,0 @@
-require File.join(File.expand_path(File.dirname(__FILE__)), '../../..', 'test_helper.rb')
-require 'rbbt/util/R'
-require 'rbbt/util/R/plot'
-
-class TestRPlot < Test::Unit::TestCase
-  def __test_ggplotgif
-
-    x = R.eval_a 'rnorm(100, 0, 1)'
-    data = TSV.setup(x, "Num~#:type=:single#:cast=:to_f")
-
-    data.add_field "Val" do |k,v|
-      k
-    end
-
-    R::GIF.eog(data, (0..10).to_a.collect{|i| 100.0 / 10**i}, <<-EOF, nil, nil, :delay => 100)
-rbbt.require('reshape')
-ggplot(data) + geom_density(bw=frame.value, aes(x=Val))
-    EOF
-
-  end
-
-  def test_gif
-
-    x = R.eval_a 'rnorm(100, 0, 1)'
-    data = TSV.setup(x, "Num~#:type=:single#:cast=:to_f")
-
-    data.add_field "Val" do |k,v|
-      k
-    end
-
-    R::GIF.eog_plot(data, (0..10).to_a.collect{|i| 100.0 / 10**i}, <<-EOF, 400, 400, :delay => 10)
-rbbt.require('reshape')
-plot(density(data$Val, bw=frame.value))
-    EOF
-
-  end
-end
-

data/test/rbbt/util/concurrency/processes/test_socket.rb

@@ -1,70 +0,0 @@
-require File.join(File.expand_path(File.dirname(__FILE__)), '../../../..', 'test_helper.rb')
-
-require 'rbbt-util'
-require 'rbbt/util/log'
-require 'rbbt/util/concurrency/processes/socket'
-
-class TestConcurrency < Test::Unit::TestCase
-  def test_socket_push_pop
-    obj1 = [1,[2,3,4]] #* 1000
-    obj2 = ["1",["2","3","4"]] #* 1000
-    obj3 = "some string"
-    obj4 = TSV.setup({1 => 1})
-
-    socket = RbbtProcessQueue::RbbtProcessSocket.new  Marshal
-    10.times do
-      socket.push(obj1)
-      socket.push(obj2)
-      socket.push(obj3)
-      socket.push(obj4)
-
-      assert_equal obj1, socket.pop
-      assert_equal obj2, socket.pop 
-      assert_equal obj3, socket.pop 
-      assert_equal obj4, socket.pop 
-
-    end
-
-    socket.swrite.close
-    assert_raise ClosedStream do 
-      socket.pop
-    end
-
-    socket.clean
-  end
-end
-
-if false and __FILE__ == $0
-  socket = RbbtProcessQueue::RbbtProcessSocket.new 
-
-  obj = "Some string" * 1000
-  Misc.benchmark(1000) do
-    socket.push(obj)
-    socket.pop
-  end
-
-  obj = ["Some string"] * 1000
-  Misc.benchmark(1000) do
-    socket.push(obj)
-    socket.pop
-  end
-  socket.clean
-
-
-  socket = RbbtProcessQueue::RbbtProcessSocket.new Marshal
-
-  obj = "Some string" * 1000
-  Misc.benchmark(1000) do
-    socket.push(obj)
-    socket.pop
-  end
-  socket.clean
-
-  socket = RbbtProcessQueue::RbbtProcessSocket.new TSV::StringArraySerializer
-  obj = ["Some string"] * 1000
-  Misc.benchmark(1000) do
-    socket.push(obj)
-    socket.pop
-  end
-  socket.clean
-end