snp-search 0.21.0 → 0.22.0

data/README.rdoc

@@ -58,7 +58,7 @@ Alternatively, you may download a SQL tool to see a GUI of your database (e.g. S
 
 We have included two example queries that you may find useful:
 
-* Example1: This script queries the database and selects all genes except the phage genes.  The output is a FASTA file of the genes.  This is a way of removing a set of genes that are not needed for the SNP analysis. You may use this script to do other SQL queries that result in a FASTA output.
+* Example1: This script queries the database to select only those SNPs not found in phage related genes. These SNPs were used to make a concatenated SNP multiple alignment file (FASTA format).  This is a way of removing a set of genes that are not needed for the SNP analysis. You may use this script to do other SQL queries that result in a FASTA output.
 
 Usage:
 

data/VERSION

@@ -1 +1 @@
-0.21.0
+0.22.0

data/bin/snp-search

@@ -10,10 +10,9 @@ opts = Slop.new :help do
   banner "ruby snp-search [OPTIONS]"
 
   on :V, :verbose, 'Enable verbose mode'
-  on :n, :name=, 'Name of database, (default: snp_db.sqlite3)', :default => 'snp_db.sqlite3' 
+  on :n, :name=, 'Name of database, Required', true 
   on :r, :reference_file=, 'Reference genome file, in gbk or embl file format, Required', true
   on :v, :vcf_file=, '.vcf file, Required', true
-  on :s, :strain=, 'text file with a list of strains/samples, Required', true
   on :c, :cuttoff_snp=, 'SNP quality cutoff, (default = 90)', :default => 90 
   on :t, :cuttoff_genotype=, 'Genotype quality cutoff (default = 30)', :default => 30
    
@@ -23,9 +22,9 @@ opts = Slop.new :help do
 end
 opts.parse
 
+  puts "You must supply the -n option, it's a required field" and exit unless opts[:name]
   puts "You must supply the -r option, it's a required field" and exit unless opts[:reference_file]
   puts "You must supply the -v option, it's a required field" and exit unless opts[:vcf_file] 
-  puts "You must supply the -s option, it's a required field" and exit unless opts[:strain]
    
   begin 
   puts "#{opts[:reference_file]} file does not exist!" and exit unless File.exist?(opts[:reference_file])
@@ -37,19 +36,7 @@ opts.parse
   rescue
   end
 
-  begin
-  puts "#{opts[:strain]} file does not exist!" and exit unless File.exist?(opts[:strain])
-  rescue
-  end
-
-
 begin
-strains = []
-  File.read(opts[:strain]).each_line do |line| 
-    strains << line.chop
-  end
- 
-
 # Enter the name of your database 
 establish_connection(opts[:name])
 
@@ -74,15 +61,11 @@ sequence_format = guess_sequence_format(ref)
 # path for vcf file here
 vcf_mpileup_file = opts[:vcf_file]
 
-
-# The populate_strains method populates the strains in the db. It uses the strain names in array.
-populate_strains(strains)
-
 # The populate_features_and_annotations method populates the features and annotations.  It uses the embl/gbk file.
 populate_features_and_annotations(sequence_flatfile)
 
-#The populate_snps_alleles_genotypes method populates the snps, alleles and genotypes.  It uses the strain names (array) and vcf file.
-populate_snps_alleles_genotypes(strains, vcf_mpileup_file, opts[:cuttoff_snp].to_i, opts[:cuttoff_genotype].to_i)
+#The populate_snps_alleles_genotypes method populates the snps, alleles and genotypes.  It uses the vcf file, and if specified, the SNP quality cutoff and genotype quality cutoff
+populate_snps_alleles_genotypes(vcf_mpileup_file, opts[:cuttoff_snp].to_i, opts[:cuttoff_genotype].to_i)
 
 rescue
 end

data/examples/example1.rb

@@ -2,9 +2,7 @@
 # Only use this script once your database has been fully populated.
 # Usage: ruby example1.rb -d your_db_name.sqlite3 -s list_of_your_species.txt -o output.fasta 
 # You may use this script to do other SQL queries that result in a fasta output.  Just change the 'snps' SQL query below with your query.
-require 'snp_db_connection'
 require 'snp_db_models'
-require 'snp_db_schema'
 gem "slop", "~> 2.4.0"
 require 'slop'
 
@@ -15,7 +13,8 @@ opts = Slop.new :help do
   on :d, :database=, 'The name of the database you like to query', true
   on :o, :outfile=, 'output file, in fasta format', true  
   on :s, :strain=, 'The strains/samples you like to query', true
-   
+  on :a, :annotation=, 'The gene you like to remove from analysis', true
+     
   on_empty do
     puts help
   end
@@ -35,6 +34,7 @@ begin
 rescue
 end
 
+annotation = opts[:annotation]
 establish_connection(opts[:database])
    
 begin
@@ -68,7 +68,7 @@ snps = Snp.find_by_sql("SELECT snps.* FROM snps
               (select distinct features.id FROM features 
               INNER JOIN annotations ON 
               annotations.feature_id = features.id 
-              WHERE annotations.value LIKE '%phage%'))") 
+              WHERE annotations.value LIKE '%(#{annotation})%'))") 
 
 
 #puts snps.size

data/examples/list_of_GAS_strains.txt

@@ -0,0 +1,212 @@
+H041200152
+H041260144
+H041320325
+H041380342
+H041520010
+H041620019
+H041680416
+H041740036
+H041860019
+H042140018
+H040300050
+H042220216
+H041400347
+H041460313
+H042320017
+H040360231
+H040640029
+H040660409
+H040680243
+H040700032
+H040960438
+H041080566
+H041120010
+H044140024
+H044300140
+H044600024
+H044760104
+H045220067
+H045220068
+H050200647
+H050620079
+H080540148
+H080580153
+H042380354
+H080680108
+H080700014
+H080920125
+H081100032
+H042560017
+H042660021
+H042880341
+H043080575
+H043280056
+H043820282
+H043920025
+H044020657
+H081880240
+H081940295
+H081980138
+H082060034
+H082140058
+H082160009
+H082160010
+H082240367
+H082320060
+H082340085
+H081120076
+H082340087
+H082400032
+H082420086
+H081180049
+H081200095
+H081220201
+H081380045
+H081420209
+H081520265
+H081520277
+H081860326
+H085140304
+H085180220
+H085180222
+H090140403
+H090140410
+H090220294
+H090240476
+H090340417
+H090360446
+H090400355
+H082500121
+H090460251
+H090480157
+H090500230
+H082660048
+H082780120
+H082800043
+H082800044
+H082980048
+H083140517
+H083200136
+H084740200
+H091280140
+H091340198
+H091300415
+H091360124
+H091380126
+H091540320
+H091680457
+H091740414
+H091760233
+H091760238
+H090580174
+H091960112
+H091960708
+H091980013
+H090640283
+H090700200
+H090780401
+H090800100
+H090920456
+H091200143
+H091220182
+H091220183
+H090480158
+H090580172
+H090600241
+H090600242
+H090640289
+H090740100
+H090940181
+H090960223
+H091300411
+H091560205
+H092120149
+H091640755
+H091680438
+H091760235
+H092260222
+H092480114
+H092520139
+H092780164
+H093980210
+H095060138
+H095160155
+H090200297
+H094540354
+H094760078
+H095080492
+H095100188
+H095240140
+H100180477
+H095260498
+H094680245
+H094560504
+H094360202
+H091780500
+H094180239
+H094160182
+H093840091
+H091820284
+H092080099
+H093040584
+H093080338
+H093180215
+H093340223
+H093420437
+H093640539
+H093420432
+H093380123
+H093360238
+H093260335
+H093200228
+H093180214
+H093120266
+H093100592
+H093080653
+H093060566
+H093780119
+H092940206
+H092940205
+H092920374
+H093700446
+H093640534
+H093560718
+H093560202
+H093520331
+H093500353
+H093440288
+H093420433
+H092560200
+H092300123
+H092300122
+H092280040
+H092260221
+H091940171
+H090200284
+H091820288
+H091780492
+H091640750
+H092920373
+H090980124
+H090900473
+H090300377
+H092920369
+H092880154
+H092880153
+H092860554
+H092860549
+H092840305
+H092800528
+H092600318
+MGAS10270
+MGAS10394
+MGAS10750
+MGAS2096
+MGAS315
+MGAS5005
+MGAS6180
+MGAS8232
+MGAS9429
+Manfredo
+NZ131
+SSI

data/examples/snp_db_models.rb

@@ -0,0 +1,32 @@
+require 'snp_db_connection'
+
+class Strain < ActiveRecord::Base
+  has_many :alleles, :through => :genotypes
+  has_many :genotypes
+end
+
+class Feature < ActiveRecord::Base
+  has_many :annotations
+  has_many :snps
+end
+
+class Snp < ActiveRecord::Base
+  belongs_to :feature
+  has_many  :alleles
+  belongs_to :reference_allele, :class_name => "Allele", :foreign_key => "reference_allele_id"
+end
+
+class Allele < ActiveRecord::Base
+  has_many :genotypes
+  belongs_to :snp
+  has_many :strains, :through => :genotypes
+end
+
+class Genotype < ActiveRecord::Base
+  belongs_to :allele
+  belongs_to :strain
+end
+
+class Annotation < ActiveRecord::Base
+  belongs_to :feature
+end

data/lib/snp-search.rb

@@ -16,15 +16,6 @@ def guess_sequence_format(reference_genome)
   return file_format
 end
 
-# A method to populate the strain names in the Strain table.  strain_names is an array of strain names.
-def populate_strains(strain_names)
-	strain_names.each do |strain|
-	s = Strain.new
-	s.name = strain
-	s.save
-	end
-end
-
 # A method to populate the database with the features (genes etc) and the annotations from the embl file.  
 # We include all features that are not 'source' or 'gene' as they are repetitive info.  'CDS' is the gene.
 # The annotation table includes also the start and end coordinates of the CDS.  The strand is also included.  the 'locations' method is defined in bioruby under genbank.  It must be required at the top (bio).
@@ -56,12 +47,8 @@ end
 #This method populates the rest of the information, i.e. SNP information, Alleles and Genotypes.
 # It requires the strain_names as array and the output (vcf file) from mpileup-snp identification algorithm.
 
-def populate_snps_alleles_genotypes(strain_names, vcf_file, cuttoff_snp, cuttoff_genotype)
-	strains = Array.new
- 	strain_names.each do |strain_name|
-  		strain = Strain.find_by_name(strain_name) # equivalent to Strain.find.where("strains.name=?", strain_name).first
-  		strains << strain
-	end
+def populate_snps_alleles_genotypes(vcf_file, cuttoff_snp, cuttoff_genotype)
+	
 
 # open vcf file and parse each line
 	File.open(vcf_file) do |f|
@@ -69,8 +56,22 @@ def populate_snps_alleles_genotypes(strain_names, vcf_file, cuttoff_snp, cuttoff
 	  header = f.gets
 	  header2 = f.gets.chomp
 	  column_headings = header2.split("\t")
-	  sample_names = column_headings[9..-1]
-	 
+	  strain_names = column_headings[9..-1]
+	 strain_names.map!{|name| name.sub(/\..*/, '')}
+	
+	strain_names.each do |str|
+		ss = Strain.new
+		ss.name = str
+		ss.save
+	end
+	
+	strains = Array.new
+ 	strain_names.each do |strain_name|
+   		strain = Strain.find_by_name(strain_name) # equivalent to Strain.find.where("strains.name=?", strain_name).first
+   		strains << strain
+	 end
+
+	  
 	  good_snps = 0
 	  # start parsing snps
 		while line = f.gets

data/snp-search.gemspec

@@ -5,7 +5,7 @@
 
 Gem::Specification.new do |s|
   s.name = "snp-search"
-  s.version = "0.21.0"
+  s.version = "0.22.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Ali Al-Shahib", "Anthony Underwood"]
@@ -28,8 +28,11 @@ Gem::Specification.new do |s|
     "Rakefile",
     "VERSION",
     "bin/snp-search",
+    "examples/ali.txt",
     "examples/example1.rb",
     "examples/example2.rb",
+    "examples/list_of_GAS_strains.txt",
+    "examples/snp_db_models.rb",
     "lib/snp-search.rb",
     "lib/snp_db_connection.rb",
     "lib/snp_db_models.rb",

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: snp-search
 version: !ruby/object:Gem::Version
-  version: 0.21.0
+  version: 0.22.0
   prerelease: 
 platform: ruby
 authors:
@@ -14,7 +14,7 @@ date: 2011-12-07 00:00:00.000000000Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activerecord
-  requirement: &2158662860 !ruby/object:Gem::Requirement
+  requirement: &2158657440 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -22,10 +22,10 @@ dependencies:
         version: 3.1.3
   type: :runtime
   prerelease: false
-  version_requirements: *2158662860
+  version_requirements: *2158657440
 - !ruby/object:Gem::Dependency
   name: bio
-  requirement: &2158662200 !ruby/object:Gem::Requirement
+  requirement: &2158656840 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -33,10 +33,10 @@ dependencies:
         version: 1.4.2
   type: :runtime
   prerelease: false
-  version_requirements: *2158662200
+  version_requirements: *2158656840
 - !ruby/object:Gem::Dependency
   name: slop
-  requirement: &2158661540 !ruby/object:Gem::Requirement
+  requirement: &2158645460 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -44,10 +44,10 @@ dependencies:
         version: 2.4.0
   type: :runtime
   prerelease: false
-  version_requirements: *2158661540
+  version_requirements: *2158645460
 - !ruby/object:Gem::Dependency
   name: sqlite3
-  requirement: &2158660960 !ruby/object:Gem::Requirement
+  requirement: &2158644880 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -55,10 +55,10 @@ dependencies:
         version: 1.3.4
   type: :runtime
   prerelease: false
-  version_requirements: *2158660960
+  version_requirements: *2158644880
 - !ruby/object:Gem::Dependency
   name: rspec
-  requirement: &2158660360 !ruby/object:Gem::Requirement
+  requirement: &2158644320 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -66,10 +66,10 @@ dependencies:
         version: 2.3.0
   type: :development
   prerelease: false
-  version_requirements: *2158660360
+  version_requirements: *2158644320
 - !ruby/object:Gem::Dependency
   name: bundler
-  requirement: &2158659780 !ruby/object:Gem::Requirement
+  requirement: &2158643800 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -77,10 +77,10 @@ dependencies:
         version: 1.0.0
   type: :development
   prerelease: false
-  version_requirements: *2158659780
+  version_requirements: *2158643800
 - !ruby/object:Gem::Dependency
   name: jeweler
-  requirement: &2158659200 !ruby/object:Gem::Requirement
+  requirement: &2158643320 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -88,10 +88,10 @@ dependencies:
         version: 1.6.4
   type: :development
   prerelease: false
-  version_requirements: *2158659200
+  version_requirements: *2158643320
 - !ruby/object:Gem::Dependency
   name: rcov
-  requirement: &2158658680 !ruby/object:Gem::Requirement
+  requirement: &2158642840 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -99,7 +99,7 @@ dependencies:
         version: '0'
   type: :development
   prerelease: false
-  version_requirements: *2158658680
+  version_requirements: *2158642840
 description: Use the snp-search toolset to query the SNP database
 email: ali.al-shahib@hpa.org.uk
 executables:
@@ -119,8 +119,11 @@ files:
 - Rakefile
 - VERSION
 - bin/snp-search
+- examples/ali.txt
 - examples/example1.rb
 - examples/example2.rb
+- examples/list_of_GAS_strains.txt
+- examples/snp_db_models.rb
 - lib/snp-search.rb
 - lib/snp_db_connection.rb
 - lib/snp_db_models.rb
@@ -143,7 +146,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
       version: '0'
       segments:
       - 0
-      hash: 2333300887326022643
+      hash: 534829390528662828
 required_rubygems_version: !ruby/object:Gem::Requirement
   none: false
   requirements: