tree_clusters 0.5.2 → 0.6.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 7d7f7982ea210cef7357e004ceb85d8b17389a32
-  data.tar.gz: a924d2bfd587ce03678c391f2a5f761848169726
+  metadata.gz: 612ebc10d0f23a15ae9d11d35be3f1de197d29fd
+  data.tar.gz: '084ff094f024a062e1c0f1476b9105950f796edf'
 SHA512:
-  metadata.gz: 7b4540a485a4635194a170c74c1bae7cd4eeb5b58df3d3a934933a161ef918903461ba75924f3baf65724050b2d3d122ac953a0822f5218c7ac0f8b37db72405
-  data.tar.gz: fced0e673bcacf6dc113d6112bd5d02b9d5fcf5756ae50d1b77cb41b1c8f97b76a3fe2ef86f8f6c7ea8ddeee96a5b5aa06600cdef310842276a486b829763eae
+  metadata.gz: 247778fdebedd3213d96ccafc52ce1d5667bdb9e8eefee60e30b3dbf1fa618ea73000d6d2a5802e47e84d9ba6fa4c119fd3acb1d21907c4c268f63e38949ffcd
+  data.tar.gz: e8d5cc7d6aa7ce86f277f42c5a5b8bd7b7d2bebe5ac3fbf5c4201315bc7aa8e411940056c30dce22226864d982083605180bfe7304c824566e107a43a308c910

data/.gitignore

@@ -23,6 +23,8 @@ TEST
 
 time.html
 
+todo.txt
+
 # rspec failure tracking
 .rspec_status
 *.lock

data/exe/snazzy_clades_attrs

@@ -0,0 +1,223 @@
+#!/usr/bin/env ruby
+
+Signal.trap("PIPE", "EXIT")
+
+require "tree_clusters"
+require "trollop"
+require "parse_fasta"
+require "shannon"
+require "fileutils"
+
+TreeClusters.extend TreeClusters
+
+def puts_info outf, clade_id, attr_cat, attr_set
+  outf.puts [clade_id, attr_cat, attr_set.to_a].join "\t"
+end
+
+opts = Trollop.options do
+  version TreeClusters::VERSION
+
+  banner <<-EOS
+
+
+  Checking IDs
+  ------------
+
+  IDs for the sequences must match between the three input files.
+
+  The tree file is allowed to have quoted taxa names, but the mapping
+  file and alignment file are not.
+
+  If your alignment file has spaces in the name, the ID part of the
+  header (i.e., the part up until the space) must match with the
+  sequence IDs in the tree and the mapping file.
+
+  Example: This would be okay.
+
+  tree file:
+    ('genome_A', 'genome_B');
+
+  aln file:
+    >genome_A apple pie
+    AAAAA
+    >genome_B brown sugar
+    AATTA
+
+  mapping file:
+    name	coolness
+    genome_A	cool
+    genome_B	notcool
+
+
+  Subtracting parent nodes
+  ------------------------
+
+  If a clade's parent would be the root of the tree, no columns will
+  be subtracted when removing the parent columns as it would be the
+  entire alignment.
+
+  Options:
+  EOS
+
+  opt(:tree,
+      "Newick tree file",
+      type: :string)
+  opt(:mapping,
+      "Mapping file",
+      type: :string)
+  opt(:attrs,
+      "Attributes file",
+      type: :string)
+
+  opt(:entropy_cutoff,
+      "Cutoff to consider a column low entropy",
+      default: 0.0)
+  opt(:clade_size_cutoff,
+      "Consider only clades with at least this many leaves",
+      default: 1)
+
+  opt(:outdir,
+      "Output directory",
+      default: ".")
+  opt(:base,
+      "Basename for output",
+      default: "snazzy_clades")
+end
+
+abort_if opts[:tree].nil?,
+         "--tree is a required arg"
+abort_if opts[:mapping].nil?,
+         "--mapping is a required arg"
+abort_if opts[:attrs].nil?,
+         "--attrs is a required arg"
+
+abort_unless_file_exists opts[:tree]
+abort_unless_file_exists opts[:mapping]
+abort_unless_file_exists opts[:attrs]
+
+# TODO check IDs when attrs is not a fasta file
+# TreeClusters.check_ids opts[:tree], opts[:mapping], opts[:attrs]
+
+abort_unless opts[:entropy_cutoff] >= 0,
+             "--entropy-cutoff must be >= 0"
+abort_unless opts[:clade_size_cutoff] >= 1,
+             "--clade-size-cutoff must be >= 1"
+
+FileUtils.mkdir_p opts[:outdir]
+
+tree                   = NewickTree.fromFile opts[:tree]
+metadata               = TreeClusters.read_mapping_file opts[:mapping]
+snazzy_clades          = TreeClusters.snazzy_clades tree, metadata
+attr_names, leaf2attrs = TreeClusters.read_attrs_file opts[:attrs]
+
+clades_fname  =
+  File.join opts[:outdir],
+            "#{opts[:base]}.snazzy_clades.txt"
+members_fname =
+  File.join opts[:outdir],
+            "#{opts[:base]}.snazzy_clades_clade_members.txt"
+attrs_fname =
+  File.join opts[:outdir],
+            "#{opts[:base]}.snazzy_clades_attrs_union.txt"
+attrs_intersection_fname =
+  File.join opts[:outdir],
+            "#{opts[:base]}.snazzy_clades_attrs_intersection.txt"
+attrs_minus_parent_attrs_fname =
+  File.join opts[:outdir],
+            "#{opts[:base]}.snazzy_clades_attrs_minus_parent_attrs.txt"
+attrs_minus_sibling_attrs_fname =
+  File.join opts[:outdir],
+            "#{opts[:base]}.snazzy_clades_attrs_minus_sibling_attrs.txt"
+attrs_minus_other_attrs_fname =
+  File.join opts[:outdir],
+            "#{opts[:base]}.snazzy_clades_attrs_minus_other_attrs.txt"
+
+
+info_f =
+  File.open(clades_fname, "w")
+clade_members_f =
+  File.open(members_fname, "w")
+attrs_f =
+  File.open(attrs_fname, "w")
+attrs_intersection_f =
+  File.open(attrs_intersection_fname, "w")
+attrs_minus_parent_attrs_f =
+  File.open(attrs_minus_parent_attrs_fname, "w")
+attrs_minus_sibling_attrs_f =
+  File.open(attrs_minus_sibling_attrs_fname, "w")
+attrs_minus_other_attrs_f =
+  File.open(attrs_minus_other_attrs_fname, "w")
+
+
+begin
+  # info is { metadata_category => metadata_tag , ... }
+  snazzy_clades.each_with_index do |(clade, info), idx|
+    assert clade.all_leaves.all? { |leaf| leaf2attrs.has_key? leaf },
+           "Not all leaves are present in the leaf2attrs hash table"
+
+    clade_id = "clade_#{idx+1}___#{clade.name}"
+
+    info_f.puts [clade_id,
+                 info.count,
+                 info.map { |pair| pair.join("|")}].join "\t"
+
+    clade_members_f.puts [clade_id,
+                          clade.all_leaves.count,
+                          clade.all_leaves].join "\t"
+
+    attr_names.each do |attr_category|
+      attrs_all_leaves =
+        leaf2attrs.attrs clade.all_leaves, attr_category
+
+      attrs_all_sibling_leaves =
+        leaf2attrs.attrs clade.all_sibling_leaves,
+                         attr_category
+      attrs_parent_leaves =
+        leaf2attrs.attrs clade.parent_leaves,
+                         attr_category
+      attrs_other_leaves =
+        leaf2attrs.attrs clade.other_leaves,
+                         attr_category
+
+      attrs_all_minus_parent =
+        attrs_all_leaves.union - attrs_parent_leaves.union
+      attrs_all_minus_sibling =
+        attrs_all_leaves.union - attrs_all_sibling_leaves.union
+      attrs_all_minus_other =
+        attrs_all_leaves.union - attrs_other_leaves.union
+
+
+      puts_info attrs_f,
+                clade_id,
+                attr_category,
+                attrs_all_leaves.union
+
+      puts_info attrs_intersection_f,
+                clade_id,
+                attr_category,
+                attrs_all_leaves.intersection
+
+      puts_info attrs_minus_parent_attrs_f,
+                clade_id,
+                attr_category,
+                attrs_all_minus_parent
+
+      puts_info attrs_minus_sibling_attrs_f,
+                clade_id,
+                attr_category,
+                attrs_all_minus_sibling
+
+      puts_info attrs_minus_other_attrs_f,
+                clade_id,
+                attr_category,
+                attrs_all_minus_other
+    end
+  end
+ensure
+  info_f.close
+  clade_members_f.close
+  attrs_f.close
+  attrs_minus_parent_attrs_f.close
+  attrs_minus_sibling_attrs_f.close
+  attrs_minus_other_attrs_f.close
+end

data/lib/tree_clusters.rb

@@ -149,11 +149,8 @@ module TreeClusters
     metadata.each do |md_cat, leaf2mdtag|
       already_checked = Set.new
       single_tag_clades = {}
-      p [md_cat, leaf2mdtag]
 
       clades.each do |clade|
-        p [clade.name, clade.all_leaves]
-
         assert clade.all_leaves.count > 1,
                "A clade cannot also be a leaf"
 
@@ -220,6 +217,45 @@ module TreeClusters
     metadata
   end
 
+  def read_attrs_file fname
+
+    attr_names = Set.new
+    File.open(fname, "rt").each_line.with_index do |line, idx|
+      unless idx.zero?
+        _, attr_name, _ = line.chomp.split "\t"
+
+        attr_names << attr_name
+      end
+    end
+
+    attr_names = attr_names.to_a.sort
+
+    attrs = TreeClusters::Attrs.new
+
+    File.open(fname, "rt").each_line.with_index do |line, idx|
+      unless idx.zero?
+        leaf, attr_name, attr_val = line.chomp.split "\t"
+
+        if attrs.has_key? leaf
+          if attrs[leaf].has_key? attr_name
+            attrs[leaf][attr_name] << attr_val
+          else
+            attrs[leaf][attr_name] = Set.new([attr_val])
+          end
+        else
+          attrs[leaf] = {}
+
+          attr_names.each do |name|
+            attrs[leaf][name] = Set.new
+          end
+          attrs[leaf][attr_name] << attr_val
+        end
+      end
+    end
+
+      [attr_names, attrs]
+  end
+
   # A Hash table for genome/leaf/taxa attributes
   class Attrs < Hash
 

data/lib/tree_clusters/version.rb

@@ -1,3 +1,3 @@
 module TreeClusters
-  VERSION = "0.5.2"
+  VERSION = "0.6.0"
 end

data/test_files/small.attrs

@@ -0,0 +1,17 @@
+name	attr name	attr val
+a-1	fruta	manzana
+a-1	fruta	pera
+a-1	color	rojo
+a-2	fruta	manzana
+a-2	color	azul
+a-2	fruta	pera
+a-2	color	rojo
+b-1	fruta	pera
+b-1	color	blanco
+b-2	color	blanco
+bb-1	color	blanco
+bbb-1	color	blanco
+bbb-2	color	blanco
+bbb-2	color	gris
+bbb-1	tamaño	pequeña
+bbb-2	tamaño	pequeña

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: tree_clusters
 version: !ruby/object:Gem::Version
-  version: 0.5.2
+  version: 0.6.0
 platform: ruby
 authors:
 - Ryan Moore
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2017-11-02 00:00:00.000000000 Z
+date: 2017-11-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -173,6 +173,7 @@ email:
 - moorer@udel.edu
 executables:
 - snazzy_clades
+- snazzy_clades_attrs
 - snazzy_clades_key_cols
 extensions: []
 extra_rdoc_files: []
@@ -188,6 +189,7 @@ files:
 - bin/console
 - bin/setup
 - exe/snazzy_clades
+- exe/snazzy_clades_attrs
 - exe/snazzy_clades_key_cols
 - lib/tree_clusters.rb
 - lib/tree_clusters/version.rb
@@ -195,6 +197,7 @@ files:
 - test_files/non_bifurcating.aln
 - test_files/non_bifurcating.tre
 - test_files/small.aln
+- test_files/small.attrs
 - test_files/small.mapping
 - test_files/small.tre
 - test_files/small_aln_bad_ids