mspire 0.2.4 → 0.3.0

data/specs/bin/filter_and_validate__multiple_vals_helper.yaml

@@ -0,0 +1,202 @@
+--- 
+pephits_precision: 
+- validator: decoy
+  value: 0.992932862190813
+- validator: badAA
+  value: 0.178006237270664
+- validator: badAA
+  value: -0.0247654296463377
+- validator: badAA
+  value: 0.301413862599215
+- validator: bias
+  value: 0.19471183136347
+- validator: bias
+  value: -4.79308077902868
+- validator: bias
+  value: 0.402836536134372
+- validator: tmm
+  value: 0.437921755087444
+- validator: tmm
+  value: -0.267185328932326
+- validator: tmm
+  value: -0.155831020815818
+- validator: tmm
+  value: -0.0201924414730414
+- validator: tps
+  value: 0.226148409893993
+pephits: 283
+prothits_precision: 
+- validator: decoy
+  values: 
+    :normal_stdev: 0.00659628206397061
+    :normal: 0.99622641509434
+    :worst: 0.971698113207547
+- validator: badAA
+  values: 
+    :normal_stdev: 0.0305695315962635
+    :normal: 0.334905660377358
+    :worst: 0.0188679245283019
+- validator: badAA
+  values: 
+    :normal_stdev: 0.0
+    :normal: 0.0
+    :worst: 0.0
+- validator: badAA
+  values: 
+    :normal_stdev: 0.0263288499231859
+    :normal: 0.497169811320755
+    :worst: 0.0377358490566038
+- validator: bias
+  values: 
+    :normal_stdev: 0.0329963984643387
+    :normal: 0.355660377358491
+    :worst: 0.0188679245283019
+- validator: bias
+  values: 
+    :normal_stdev: 0.0
+    :normal: 0.0
+    :worst: 0.0
+- validator: bias
+  values: 
+    :normal_stdev: 0.0266832747617573
+    :normal: 0.613207547169811
+    :worst: 0.0943396226415094
+- validator: tmm
+  values: 
+    :normal_stdev: 0.0281442148871296
+    :normal: 0.65377358490566
+    :worst: 0.122641509433962
+- validator: tmm
+  values: 
+    :normal_stdev: 0.0
+    :normal: 0.0
+    :worst: 0.0
+- validator: tmm
+  values: 
+    :normal_stdev: 0.0
+    :normal: 0.0
+    :worst: 0.0
+- validator: tmm
+  values: 
+    :normal_stdev: 0.0
+    :normal: 0.0
+    :worst: 0.0
+- validator: tps
+  values: 
+    :normal_stdev: 0.0317750140353944
+    :normal: 0.402830188679245
+    :worst: 0.0283018867924528
+params: 
+  validators: 
+  - :type: decoy
+    :class: Validator::Decoy
+    :constraint: /^DECOY_/
+    :decoy_on_match: true
+    :correct_wins: true
+  - :calculated_background: 0.127208480565371
+    :type: badAA
+    :class: Validator::AA
+    :background: 0.001
+    :frequency: 0.0147528119278054
+    :false_to_total_ratio: 1.0
+  - :calculated_background: 0.402826855123675
+    :type: badAA
+    :class: Validator::AA
+    :background: 0.0
+    :frequency: 0.0463510332199843
+    :false_to_total_ratio: 1.0
+  - :calculated_background: 0.127208480565371
+    :type: badAA
+    :class: Validator::AA
+    :background: 0.001
+    :frequency: 
+    :false_to_total_ratio: 0.180662732637313
+  - :calculated_background: 0.773851590106007
+    :type: bias
+    :class: Validator::Bias
+    :proteins_expected: true
+    :correct_wins: true
+    :background: 0.0
+    :file: /work/john/mspire/specs/../test_files/validator_hits_separate/bias_bioworks_small_HS.fasta
+    :false_to_total_ratio: 0.960962324103495
+  - :calculated_background: 0.226148409893993
+    :type: bias
+    :class: Validator::Bias
+    :proteins_expected: false
+    :correct_wins: true
+    :background: 0.0
+    :file: /work/john/mspire/specs/../test_files/validator_hits_separate/bias_bioworks_small_HS.fasta
+    :false_to_total_ratio: 0.0390376758965048
+  - :calculated_background: 0.773851590106007
+    :type: bias
+    :class: Validator::Bias
+    :proteins_expected: true
+    :correct_wins: true
+    :background: 0.2
+    :file: /work/john/mspire/specs/../test_files/validator_hits_separate/bias_bioworks_small_HS.fasta
+    :false_to_total_ratio: 0.960962324103495
+  - :calculated_background: 0.359430604982206
+    :type: tmm
+    :soluble_fraction: true
+    :class: Validator::Transmem::Protein
+    :no_include_tm_peps: 0.8
+    :correct_wins: true
+    :background: 0.2
+    :transmem_file: /work/john/mspire/specs/../test_files/bioworks_small.phobius
+    :false_to_total_ratio: 0.283644859813084
+    :min_num_tms: 1
+  - :calculated_background: 0.359430604982206
+    :type: tmm
+    :soluble_fraction: true
+    :class: Validator::Transmem::Protein
+    :no_include_tm_peps: 0.8
+    :correct_wins: true
+    :background: 0.0
+    :transmem_file: /work/john/mspire/specs/../test_files/bioworks_small.phobius
+    :false_to_total_ratio: 0.283644859813084
+    :min_num_tms: 1
+  - :calculated_background: 0.293286219081272
+    :type: tmm
+    :soluble_fraction: true
+    :class: Validator::Transmem::Protein
+    :no_include_tm_peps: false
+    :correct_wins: true
+    :background: 0.0
+    :transmem_file: /work/john/mspire/specs/../test_files/bioworks_small.toppred.out
+    :false_to_total_ratio: 0.253744893327281
+    :min_num_tms: 3
+  - :calculated_background: 0.472924187725632
+    :type: tmm
+    :soluble_fraction: true
+    :class: Validator::Transmem::Protein
+    :no_include_tm_peps: 0.8
+    :correct_wins: true
+    :background: 0.0
+    :transmem_file: /work/john/mspire/specs/../test_files/bioworks_small.toppred.out
+    :false_to_total_ratio: 0.463563704748472
+    :min_num_tms: 1
+  - :type: tps
+    :class: Validator::TruePos
+  ties: true
+  digestion: 
+  - /work/john/mspire/specs/../test_files/bioworks_small.fasta
+  - /work/john/mspire/specs/../test_files/bioworks_small.params
+  prefilter: false
+  output: 
+  - text_table => /work/john/mspire/specs/../test_files/table_output.tmp
+  - yaml => /work/john/mspire/specs/../test_files/filter_and_validate.tmp
+  sequest: 
+    include_deltacnstar: true
+    xcorr1: 0.0
+    ppm: 1000000.0
+    xcorr2: 0.0
+    deltacn: 0.01
+    xcorr3: 0.0
+  top_hit_by: xcorr
+  decoy_on_match: true
+  postfilter: top_per_scan
+  include_ties_in_top_hit_postfilter: false
+  hits_together: true
+  proteins: true
+  include_ties_in_top_hit_prefilter: true
+prothits: 106

data/specs/bin/filter_and_validate_spec.rb

@@ -0,0 +1,124 @@
+require File.expand_path( File.dirname(__FILE__) + '/../spec_helper' )
+
+require 'spec_id/precision/filter'
+
+describe 'filter_and_validate.rb on small bioworks file' do
+  before(:all) do
+    @progname = 'filter_and_validate.rb'
+
+    @outfile = Tfiles + '/filter_and_validate.tmp'
+
+    # direct call with an array
+    @direct_call = Proc.new {|ar| SpecID::Precision::Filter.new.filter_and_validate_cmdline(ar) }
+    # direct call with a string
+    @direct_call_st = Proc.new {|st| @direct_call.call(st.split(/\s+/)) }
+    @st_to_yaml = Proc.new do |st| 
+      to_call = st + " -o yaml:#{@outfile} "
+      @direct_call.call(to_call.split(/\s+/)) 
+      YAML.load_file(@outfile)
+    end
+
+    @args = ["-1 0.6 -2 0.8 -3 0.9 -d 0.2", (Tfiles + '/bioworks_small.xml ')].join(' ')
+    @interactive_file = Tfiles + '/interactive.tmp'
+    File.open(@interactive_file,'w') do |fh| 
+      string = ["0.6 0.8 0.9 0.2 5000", "dcns:f", "0.6 0.8 dcns:t", "pf:s", "pf:ac", "pf:a"].join("\n")
+      fh.puts string
+    end
+    # uses DECOY_ prefix on two
+    @fake_bioworks_file = Tfiles + '/validator_hits_separate/bioworks_small_HS.xml'
+    @small_bias_fasta_file = Tfiles + '/validator_hits_separate/bias_bioworks_small_HS.fasta'
+    @small_fasta_file = Tfiles + '/bioworks_small.fasta'
+    @params_file = Tfiles + '/bioworks_small.params'
+    @toppred_file = Tfiles + '/bioworks_small.toppred.out'
+    @phobius_file = Tfiles + '/bioworks_small.phobius'
+    @table_output_file = Tfiles + '/table_output.tmp'
+  end
+
+  after(:all) do
+    [@outfile, @interactive_file, @table_output_file].each do |file|
+      File.unlink(file) if File.exist?(file)
+    end
+  end
+
+  # this ensures that the actual commandline version gives usage.
+  it_should_behave_like "a cmdline program"
+
+  it 'filters a file and outputs to table or yaml' do
+    @direct_call_st.call( @args + " -o text_table:#{@outfile}")
+    IO.read(@outfile).should =~ /66/
+    struct = @st_to_yaml.call( @args )
+    struct['pephits'].should == 66
+  end
+
+  it 'responds to --no_deltacnstar' do
+    reply_without = @st_to_yaml.call( @args + " --no_deltacnstar" )
+    reply_without['pephits'].should == 34
+  end
+
+  it 'works with interactive input (includes dcnstar and postfilter)' do
+    @direct_call_st.call( "-o text_table:#{@outfile} -i #{@interactive_file} "  + Tfiles + '/bioworks_small.xml ' )
+    reply = IO.read(@outfile)
+
+    exp = %w(73 40 73 73 33 33)
+    reply.scan(/^peps\s+(\d+)/) do |v|
+      Regexp.last_match[1] == exp.shift
+    end
+  end
+
+  it 'responds to ppm filter' do
+    reply_without = @st_to_yaml.call( @args + " -p 280" )
+    reply_without['pephits'].should == 11
+  end
+
+  it 'responds to --hits_separate' do
+    # this file has two decoy peps that score better than the real peps at
+    # those scans
+        ht_file = Tfiles + '/test_together.tmp.yaml'
+    hs_file = Tfiles + '/test_separate.tmp.yaml'
+    outputs = [ht_file, hs_file].zip(['', ' --hits_separate']).map do |output_file, flag|
+      run_normal = @cmd + " --bias #{@small_bias_fasta_file} --decoy /^DECOY_/ --digestion #{@small_fasta_file},#{@params_file} #{@fake_bioworks_file} -1 0.0 -2 0.0 -3 0.0 -d 0.01 -p 1000000 -o yaml:#{output_file} #{flag}"
+      `#{run_normal}`
+    end
+    structs = [ht_file, hs_file].map do |file| 
+      file.should exist
+      struct = YAML.load_file(file)
+      File.unlink file
+      struct
+    end
+
+    comparisons = %w(precision calc_bkg hits_together_param)
+    comps = structs.map do |st|
+      # note that calculated_background may need to be a string if we get our
+      # act together...
+      [ st['pephits_precision'][0]['value'], st['params']['validators'][0][:calculated_background], st['params']['hits_together'] ]
+    end
+    comparisons.zip( *comps ) do |tp, ht, hs|
+      ht.should_not == hs
+    end
+
+  end
+
+  it 'raises error on > 1 decoy validator' do
+    lambda { @st_to_yaml.call( @args + " --decoy /hello/ --decoy path/to/file" ) }.should raise_error(ArgumentError)
+  end
+
+  it 'handles multiple validators of the same kind (except, of course, decoy)' do
+
+    struct = @st_to_yaml.call( "#{@fake_bioworks_file} --proteins  -1 0.0 -2 0.0 -3 0.0 -d 0.01 -p 1000000 --decoy /^DECOY_/ --digestion #{@small_fasta_file},#{@params_file} --bad_aa C,true,0.001 --bad_aa E,true --bad_aa C,false,0.001 --bias #{@small_bias_fasta_file},true --bias #{@small_bias_fasta_file},false --bias #{@small_bias_fasta_file},true,0.2 --tmm #{@phobius_file},1,true,0.8,0.2 --tmm #{@phobius_file} --tmm #{@toppred_file},3,true,false --tmm #{@toppred_file} --tps #{@small_bias_fasta_file} -o text_table:#{@table_output_file} " )
+    frozen = YAML.load_file( File.dirname(__FILE__) + "/filter_and_validate__multiple_vals_helper.yaml" )
+    struct.should == frozen
+
+    text_table = IO.read(@table_output_file)
+
+    # frozen
+    headings_re = Regexp.new( %w(num decoy badAA badAA badAA  bias  bias  bias   tmm   tmm   tmm   tmm   tps).join("\\s+") )
+    data_re = Regexp.new( %w(peps 283 0.993 0.178 -0.025 0.301 0.195 -4.793 0.403 0.438 -0.267 -0.156 -0.020 0.226).join("\\s+") )
+    prot_re = Regexp.new( %w(106 0.972 0.019   0.0 0.038 0.019   0.0 0.094 0.123   0.0   0.0   0.0 0.028).join("\\s+") )
+    text_table.should =~ headings_re
+    text_table.should =~ data_re
+    text_table.should =~ prot_re
+  end
+
+end
+
+

data/specs/bin/ms_to_lmat_spec.rb

@@ -0,0 +1,34 @@
+require File.expand_path( File.dirname(__FILE__) + '/../spec_helper' )
+
+
+describe 'ms_to_lmat.rb' do
+
+  before(:all) do
+    @progname = 'ms_to_lmat.rb'
+    @mzxml = Tfiles + "/opd1/twenty_scans.mzXML"
+    @ans_lmata = Tfiles + "/opd1/twenty_scans_answ.lmata"
+    @ans_lmat = Tfiles + "/opd1/twenty_scans_answ.lmat"
+  end
+
+  it_should_behave_like "a cmdline program"
+
+  it 'creates the correct lmata (ascii) file' do
+    cmd = "#{@cmd} #{@mzxml} --ascii"   
+    `#{cmd}`
+    newfile = @mzxml.sub(".mzXML", ".lmata")
+    newfile.should exist
+    IO.read(newfile).should == IO.read(@ans_lmata)
+    File.unlink(newfile)
+  end
+
+
+  it 'creates the correct lmat (binary) file' do
+    cmd = "#{@cmd} #{@mzxml}"   
+    `#{cmd}`
+    newfile = @mzxml.sub(".mzXML", ".lmat")
+    newfile.should exist
+    IO.read(newfile).should == IO.read(@ans_lmat)
+    File.unlink(newfile)
+  end
+end
+

data/specs/bin/prob_validate_spec.rb

@@ -0,0 +1,62 @@
+require File.expand_path( File.dirname(__FILE__) + '/../spec_helper' )
+
+require 'spec_id/precision/prob'
+
+describe 'filter_and_validate.rb on small bioworks file' do
+  before(:all) do
+    @progname = 'prob_validate.rb'
+
+    @outfile = Tfiles + '/prob_and_validate.tmp'
+
+    # direct call with an array
+    @direct_call = Proc.new {|ar| SpecID::Precision::Prob.new.precision_vs_num_hits_cmdline(ar) }
+    # direct call with a string
+    @direct_call_st = Proc.new {|st| @direct_call.call(st.split(/\s+/)) }
+    @st_to_yaml = Proc.new do |st| 
+      to_call = st + " -o yaml:#{@outfile} "
+      @direct_call.call(to_call.split(/\s+/)) 
+      YAML.load_file(@outfile)
+    end
+
+    file = Tfiles + '/opd1/000_020_3prots-prot.mod_initprob.xml'
+    @args = [file].join(' ')
+    # uses DECOY_ prefix on two
+    @fake_bioworks_file = Tfiles + '/validator_hits_separate/bioworks_small_HS.xml'
+    @small_bias_fasta_file = Tfiles + '/validator_hits_separate/bias_bioworks_small_HS.fasta'
+    @small_fasta_file = Tfiles + '/bioworks_small.fasta'
+    @params_file = Tfiles + '/bioworks_small.params'
+    @toppred_file = Tfiles + '/bioworks_small.toppred.out'
+    @phobius_file = Tfiles + '/bioworks_small.phobius'
+  end
+
+  after(:all) do 
+    [@outfile].each do |file|
+      File.unlink(file) if File.exist?(file)
+    end
+  end
+
+  # this ensures that the actual commandline version gives usage.
+  it_should_behave_like "a cmdline program"
+
+  it 'outputs to yaml' do
+    reply = @st_to_yaml.call( @args )
+    keys = [:probabilities, :params, :pephits_precision, :charges, :aaseqs, :count].map {|v| v.to_s }.sort
+    reply.keys.map {|v| v.to_s}.sort.should == keys
+  end
+
+  it 'responds to --prob init' do
+    normal = @st_to_yaml.call( @args + " --prob" )
+    normal[:pephits_precision].first[:values].should == [1.0, 1.0, 0.996655518394649, 0.918918918918919]
+    #normal_nsp = @st_to_yaml.call( @args + " --prob nsp" )
+    #normal.should == normal_nsp
+    init = @st_to_yaml.call( @args + " --prob init" )
+    init.should_not == normal
+    init[:pephits_precision].first[:values].should == [1.0, 0.974358974358974, 0.981324278438031, 0.890429958391123]
+    with_sort_by = @st_to_yaml.call( @args + " --prob nsp --sort_by_init" )
+    # frozen
+    with_sort_by[:pephits_precision].first[:values].should == [1.0, 0.994974874371859, 0.996655518394649, 0.918918918918919]
+  end
+
+end
+
+

data/specs/bin/protein_summary_spec.rb

@@ -0,0 +1,10 @@
+require File.expand_path( File.dirname(__FILE__) + '/../spec_helper' )
+
+xdescribe 'protein_summary.rb' do
+
+   before(:all) do
+    @progname = 'protein_summary.rb'
+  end
+  it_should_behave_like 'a cmdline program'
+
+end