exe/key_cols in tree_clusters-0.8.2 vs exe/key_cols in tree_clusters-0.8.3

@@ -1,7 +1,9 @@
 #!/usr/bin/env ruby
 
+# TODO the block variables called "clade" aren't actually instances of the TreeClusters::Clade class.  It's kind of confusing.
+
 Signal.trap("PIPE", "EXIT")
 
 require "tree_clusters"
 require "trollop"
 require "parse_fasta"
@@ -52,10 +54,12 @@
 
   Hi.  My name is key_cols.  If you give me a Newick tree file and an
   alignment file (fasta format), I will tell you key columns for all
   clades/clusters that have them.
 
+  Version: v#{TreeClusters::VERSION}
+
 Overview
 --------
 
   A clade has key columns if you can use the residue/nucleotide at
   those columns to tell sequences in the clade from sequences outside
@@ -145,13 +149,14 @@
 abort_unless opts[:clade_size_cutoff] >= 1,
              "--clade-size-cutoff must be >= 1"
 
 FileUtils.mkdir_p opts[:outdir]
 
-AbortIf.logger.info { "Parsing input files" }
+AbortIf.logger.info { "Reading tree" }
+tree = NewickTree.fromFile opts[:tree]
 
-tree       = NewickTree.fromFile opts[:tree]
+AbortIf.logger.info { "Reading alignment" }
 leaf2attrs = TreeClusters.read_alignment opts[:aln]
 
 members_fname        =
     File.join opts[:outdir],
               "#{opts[:base]}.tree_clusters.clade_members.txt"
@@ -167,26 +172,41 @@
 key_cols_f       =
     File.open(key_cols_fname, "w")
 annotated_tree_f =
     File.open(annotated_tree_fname, "w")
 
-key_col_sets       = {}
-clade_sizes        = {}
-clade_count        = TreeClusters.all_clades(tree).count
+key_col_sets = {}
+clade_sizes  = {}
+# AbortIf.logger.info { "Counting clades" }
+# clade_count        = TreeClusters.all_clades(tree).count
 change_these_names = Set.new
 
 def all_clades_helper tree
-  TreeClusters.all_clades(tree).sort_by { |cl| cl.all_leaves.count }.reverse
+  # TreeClusters.all_clades(tree).sort_by { |cl| cl.all_leaves.count }.reverse
+  tree.clade_nodes.
+      reverse.
+      sort_by { |node| node.all_leaves.count }.
+      reverse
 end
 
-AbortIf.logger.info { "Processing clades" }
+AbortIf.logger.info { "Getting all clades" }
+clades      = all_clades_helper(tree)
+clade_count = clades.count
+
+AbortIf.logger.info { "Processing clades (The first few clades go reaaally slowly, but then I speed up!)" }
 begin
-  all_clades_helper(tree).each_with_index do |clade, idx|
-    if ((idx + 1) % 100).zero?
+  clades.each_with_index do |clade, idx|
+    # It starts off really slowly, then speeds up a lot.
+    if (idx+1) < 100 ||
+        ((idx+1) < 1000 && ((idx + 1) % 10).zero?) ||
+        ((idx+1) < 10000 && ((idx + 1) % 100).zero?) ||
+        ((idx+1) < 100000 && ((idx + 1) % 1000).zero?) ||
+        ((idx + 1) % 10000).zero?
+
       perc = ((idx + 1) / clade_count.to_f * 100).round 2
 
-      STDERR.printf("Processing clades: #{perc}%\r")
+      STDERR.printf("Processing clades: #{idx + 1} of #{clade_count} (#{perc}%%)\r")
     end
 
     clade_id              = "clade_#{idx + 1}___#{clade.name.tr("'", "_")}"
     clade_sizes[clade_id] = clade.all_leaves.count
 
@@ -222,15 +242,15 @@
     end
   end
 
   AbortIf.logger.info { "Annotating tree" }
 
-  all_clades_helper(tree).each_with_index do |clade, idx|
+  clades.each_with_index do |clade, idx|
     clade_id = "clade_#{idx + 1}___#{clade.name.tr("'", "_")}"
 
     if change_these_names.include? clade_id
       # This will change the node in the original NewickTree
-      clade.node.name = "'#{clade_id}'"
+      clade.name = "'#{clade_id}'"
     end
   end
 
   annotated_tree_f.puts tree.to_s.sub(/;+$/, ";")
 ensure