miga-base 0.2.0.9 → 0.2.1.0

data/scripts/clade_finding.bash

@@ -11,33 +11,35 @@ date "+%Y-%m-%d %H:%M:%S %z" > "miga-project.start"
 
 # Markov-cluster genomes by ANI
 gunzip -c ../../09.distances/03.ani/miga-project.txt.gz | tail -n+2 \
-   | awk -F"\\t" '{print $2"'"\\t"'"$3"'"\\t"'"$4}' > genome-genome.aai90.rbm
+  | awk -F"\\t" '$4>=90{print $2"'"\\t"'"$3"'"\\t"'"$4}' \
+  > genome-genome.aai90.rbm
 ogs.mcl.rb -d . -o miga-project.aai90-clades -t "$CORES" -i \
-   -f "(\\S+)-(\\S+)\\.aai90\\.rbm"
+  -f "(\\S+)-(\\S+)\\.aai90\\.rbm"
 rm genome-genome.aai90.rbm
-cat genome-genome.aai90.rbm | awk -F"\\t" '$3>=95' > genome-genome.ani95.rbm
+gunzip -c ../../09.distances/02.aai/miga-project.txt.gz | tail -n+2 \
+  | awk -F"\\t" '$4>=95{print $2"'"\\t"'"$3"'"\\t"'"$4}' \
+  > genome-genome.ani95.rbm
 ogs.mcl.rb -d . -o miga-project.ani95-clades -t "$CORES" -b \
-   -f "(\\S+)-(\\S+)\\.ani95\\.rbm"
+  -f "(\\S+)-(\\S+)\\.ani95\\.rbm"
 rm genome-genome.ani95.rbm
 
 # Propose clade projects
 cat miga-project.ani95-clades | tail -n +2 | tr "," "\\t" | awk 'NF >= 5' \
-   > miga-project.proposed-clades
+  > miga-project.proposed-clades
 
-# Run R code
-echo "
-source('$MIGA/utils/subclades.R');
-subclades('../../09.distances/02.aai/miga-project.txt.gz',
-   'miga-project', $CORES);
-" | R --vanilla
-mv miga-project.ani.nwk miga-project.aai.nwk
-
-# Compile
-ruby "$MIGA/utils/subclades-compile.rb" . \
-   >  miga-project.class.tsv \
-   2> miga-project.class.nwk
+# Run R code (except in projects type clade)
+if [[ $(miga project_info -P "$PROJECT" -m type) != "clade" ]] ; then
+  $MIGA/utils/subclades.R \
+    ../../09.distances/02.aai/miga-project.txt.gz \
+    miga-project $CORES
+  mv miga-project.nwk miga-project.aai.nwk
+  
+  # Compile
+  ruby "$MIGA/utils/subclades-compile.rb" . \
+    >  miga-project.class.tsv \
+    2> miga-project.class.nwk
+fi
 
 # Finalize
 date "+%Y-%m-%d %H:%M:%S %z" > "miga-project.done"
 miga add_result -P "$PROJECT" -r clade_finding
-

data/scripts/distances.bash

@@ -17,7 +17,8 @@ NOMULTI=$(miga list_datasets -P "$PROJECT" -D "$DATASET" --no-multi \
 REF=$(miga list_datasets -P "$PROJECT" -D "$DATASET" --ref \
    | wc -l | awk '{print $1}')
 
-# Call submodule
+# Call submodules
+source "$MIGA/scripts/_distances_functions.bash"
 if [[ "$NOMULTI" -eq "1" && "$REF" -eq "1" ]] ; then
    source "$MIGA/scripts/_distances_ref_nomulti.bash"
 elif [[ "$NOMULTI" -eq "1" ]] ; then

data/scripts/init.bash

@@ -106,11 +106,7 @@ echo "export PATH=\$MIGA_PATH\$PATH" >> "$HOME/.miga_rc"
 # Check for R packages
 echo "
 Looking for R packages:" >&2
-if ! check_rlib enveomics.R ; then
-   echo "+ Installing enveomics.R" >&2
-   R CMD INSTALL $(dirname "$(which "FastQ.tag.rb")")/../enveomics.R
-fi
-RLIBS="ape ggdendro ggplot2 gridExtra cluster dendextend vegan scatterplot3d"
+RLIBS="enveomics.R ape ggdendro ggplot2 gridExtra cluster dendextend vegan scatterplot3d"
 for lib in $RLIBS ; do
    if ! check_rlib $lib ; then
       echo "+ Installing $lib" >&2
@@ -122,7 +118,7 @@ done
 # Check for ruby gems
 echo "
 Looking for Ruby gems:" >&2
-GEMS="rest_client sqlite3 daemons json"
+GEMS="rest-client sqlite3 daemons json"
 for gem in $GEMS ; do
    if ! check_gem $gem ; then
       echo "+ Installing $gem (user-only)" >&2

data/scripts/subclades.bash

@@ -10,11 +10,10 @@ cd "$PROJECT/data/10.clades/02.ani"
 date "+%Y-%m-%d %H:%M:%S %z" > "miga-project.start"
 
 # Run R code
-echo "
-source('$MIGA/utils/subclades.R');
-subclades('../../09.distances/03.ani/miga-project.txt.gz',
-   'miga-project', $CORES);
-" | R --vanilla
+$MIGA/utils/subclades.R \
+  ../../09.distances/03.ani/miga-project.txt.gz \
+  miga-project $CORES
+mv miga-project.nwk miga-project.ani.nwk
 
 # Compile
 ruby "$MIGA/utils/subclades-compile.rb" . \

data/test/common_test.rb

@@ -3,7 +3,7 @@ require "test_helper"
 class CommonTest < Test::Unit::TestCase
   
   def setup
-    $jruby_tests = !ENV["JRUBY_TESTS"].nil?
+    #$jruby_tests = !ENV["JRUBY_TESTS"].nil?
   end
 
   def test_debug
@@ -28,7 +28,7 @@ class CommonTest < Test::Unit::TestCase
     assert_respond_to(MiGA::MiGA, :DEBUG)
     assert_respond_to(MiGA::MiGA, :DEBUG_ON)
     assert_respond_to(MiGA::MiGA, :DEBUG_OFF)
-    omit_if($jruby_tests, "JRuby doesn't like interceptions.")
+    #omit_if($jruby_tests, "JRuby doesn't like interceptions.")
     MiGA::MiGA.DEBUG_TRACE_ON
     err = capture_stderr do
       MiGA::MiGA.DEBUG "Dandadi"

data/test/daemon_test.rb

@@ -4,11 +4,13 @@ require "miga/daemon"
 class DaemonTest < Test::Unit::TestCase
   
   def setup
+    $jruby_tests = !ENV["JRUBY_TESTS"].nil?
     $tmp = Dir.mktmpdir
     ENV["MIGA_HOME"] = $tmp
     FileUtils.touch("#{ENV["MIGA_HOME"]}/.miga_rc")
     File.open("#{ENV["MIGA_HOME"]}/.miga_daemon.json", "w") do |fh|
-      fh.puts '{"maxjobs":1,"ppn":1,"latency":2}'
+      fh.puts '{"maxjobs":1,"ppn":1,"latency":2,"varsep":" ","var":"%s=%s",
+        "cmd":"%5$s","alive":"echo 1 # %s"}'
     end
     $p1 = MiGA::Project.new(File.expand_path("project1", $tmp))
     $d1 = MiGA::Daemon.new($p1)
@@ -18,7 +20,77 @@ class DaemonTest < Test::Unit::TestCase
     FileUtils.rm_rf $tmp
     ENV["MIGA_HOME"] = nil
   end
+
+  def test_check_project
   
+  end
+
+  def test_check_datasets
+    p = $p1
+    d = $d1
+    d.runopts(:maxjobs, 0, true)
+    assert(d.jobs_to_run.empty?)
+    ds = p.add_dataset("ds1")
+    d.check_datasets
+    assert(d.jobs_to_run.empty?)
+    FileUtils.cp(File.expand_path("daemon/daemon.json", p.path),
+      File.expand_path("data/01.raw_reads/ds1.1.fastq", p.path))
+    FileUtils.cp(File.expand_path("daemon/daemon.json", p.path),
+      File.expand_path("data/01.raw_reads/ds1.done", p.path))
+    out = capture_stdout do
+      d.check_datasets
+    end
+    assert(out.string =~ /Queueing #{ds.name}:trimmed_reads/)
+    assert_equal(1, d.jobs_to_run.size)
+    assert_equal("project1:trimmed_reads:ds1", d.jobs_to_run.first[:cmd])
+    assert_equal(d.jobs_to_run.first, d.get_job(:trimmed_reads, ds))
+  end
+
+  def test_in_loop
+    p = $p1
+    d = $d1
+    d.runopts(:latency, 0, true)
+    assert_equal(-1, d.loop_i)
+    assert_nil(d.last_alive)
+    out = capture_stdout do
+      d.in_loop
+    end
+    assert_equal(DateTime, d.last_alive.class)
+    assert(out.string =~ /-{20}\n.*MiGA:#{p.name} launched/)
+    10.times{ d.in_loop }
+    assert_equal(11, d.loop_i)
+    out = capture_stdout do
+      d.in_loop
+    end
+    assert(out.string =~ /Housekeeping for sanity/)
+    assert_equal(0, d.loop_i)
+  end
+
+  def test_start
+    p = $p1
+    d = $d1
+    d.runopts(:latency, 0, true)
+    assert_equal(0, d.latency)
+    omit_if($jruby_tests, "JRuby doesn't implement fork.")
+    $child = fork { d.start }
+    sleep(3)
+    dpath = File.expand_path("daemon/MiGA:#{p.name}",p.path)
+    assert(File.exist?("#{dpath}.pid"))
+    out = capture_stdout { d.stop }
+    assert(out.string =~ /MiGA:#{p.name}: trying to stop process with pid \d+/)
+    assert(!File.exist?("#{dpath}.pid"))
+    assert(File.exist?("#{dpath}.output"))
+    File.open("#{dpath}.output", "r") do |fh|
+      l = fh.each_line.to_a
+      assert(l[0] =~ /-{20}\n/)
+      assert(l[1] =~ /MiGA:#{p.name} launched/)
+      assert(l[2] =~ /-{20}\n/)
+      assert(l[3] =~ /Housekeeping for sanity\n/)
+    end
+  ensure
+    Process.kill("KILL", $child) unless $child.nil?
+  end
+
   def test_last_alive
     p = MiGA::Project.new(File.expand_path("last_alive", $tmp))
     d = MiGA::Daemon.new(p)

data/test/project_test.rb

@@ -22,7 +22,8 @@ class ProjectTest < Test::Unit::TestCase
   end
 
   def test_create
-    assert_equal($tmp + "create", MiGA::Project.new($tmp + "create").path)
+    assert_equal("#{$tmp}/create", MiGA::Project.new("#{$tmp}/create").path)
+    assert(Dir.exist?("#{$tmp}/create"))
     assert_raise do
       ENV["MIGA_HOME"] = $tmp + "/chez-moi"
       MiGA::Project.new($tmp + "/cuckoo")
@@ -41,14 +42,37 @@ class ProjectTest < Test::Unit::TestCase
   end
 
   def test_datasets
-    p = MiGA::Project.new(File.expand_path("datasets", $tmp))
+    p = $p1
     d = p.add_dataset("d1")
     assert_equal(MiGA::Dataset, d.class)
     assert_equal([d], p.datasets)
+    assert_equal(["d1"], p.dataset_names)
     p.each_dataset{ |ds| assert_equal(d, ds) }
     dr = p.unlink_dataset("d1")
     assert_equal(d, dr)
     assert_equal([], p.datasets)
+    assert_equal([], p.dataset_names)
+  end
+
+  def test_import_dataset
+    p1 = $p1
+    d1 = p1.add_dataset("d1")
+    File.open("#{p1.path}/data/01.raw_reads/#{d1.name}.1.fastq",
+      "w") { |f| f.puts ":-)" }
+    File.open("#{p1.path}/data/01.raw_reads/#{d1.name}.done",
+      "w") { |f| f.puts ":-)" }
+    d1.next_preprocessing(true)
+    p2 = MiGA::Project.new(File.expand_path("import_dataset", $tmp))
+    assert(p2.datasets.empty?)
+    assert_nil(p2.dataset("d1"))
+    p2.import_dataset(d1)
+    assert_equal(1, p2.datasets.size)
+    assert_equal(MiGA::Dataset, p2.dataset("d1").class)
+    assert_equal(1, p2.dataset("d1").results.size)
+    assert(File.exist?(
+      File.expand_path("data/01.raw_reads/#{d1.name}.1.fastq", p2.path)))
+    assert(File.exist?(
+      File.expand_path("metadata/#{d1.name}.json", p2.path)))
   end
 
 end

data/test/taxonomy_test.rb

@@ -33,4 +33,14 @@ class TaxonomyTest < Test::Unit::TestCase
     assert(tx.is_in? MiGA::Taxonomy.new("species:v3_0"))
   end
 
+  def test_init_methods
+    tx = MiGA::Taxonomy.new({:k=>"Mascot", :c=>"Cereal", :s=>"Melvin"})
+    assert_equal("k:Mascot c:Cereal s:Melvin", tx.to_s)
+    tx = MiGA::Taxonomy.new("Mascot College Buzz", "k c s")
+    assert_equal("k:Mascot c:College s:Buzz", tx.to_s)
+    assert_raise do
+      tx = MiGA::Taxonomy.new("Mascot State Georgia Peach", "k c s")
+    end
+  end
+
 end

data/test/test_helper.rb

@@ -1,5 +1,5 @@
 require "codeclimate-test-reporter"
-CodeClimate::TestReporter.start
+CodeClimate::TestReporter.start unless ENV["REMOTE_TESTS"].nil?
 
 require "rubygems"
 require "test/unit"

data/utils/subclades-compile.rb

@@ -9,7 +9,9 @@ $:.push File.expand_path(File.dirname(__FILE__) + "/lib")
 dir = ARGV.shift or abort "Usage: #{$0} <classif.dir>"
 
 def read_classif(dir, classif={})
-   fh = File.open(File.expand_path("miga-project.1.classif", dir), "r")
+   classif_file = File.expand_path("miga-project.classif", dir)
+   return classif unless File.exist? classif_file
+   fh = File.open(classif_file, "r")
    klass = []
    while ln = fh.gets
       r = ln.chomp.split("\t")
@@ -19,7 +21,7 @@ def read_classif(dir, classif={})
    end
    fh.close
    klass.each do |i|
-      d = File.expand_path("miga-project.1.sc-#{i}", dir)
+      d = File.expand_path("miga-project.sc-#{i}", dir)
       classif = read_classif(d, classif) if Dir.exist? d
    end
    classif

data/utils/subclades.R

@@ -1,171 +1,153 @@
-library(enveomics.R)
-library(ape)
-library(ggdendro)
-library(ggplot2)
-library(grid)
-library(gridExtra)
-library(cluster)
-library(dendextend)
-library(vegan)
-library(scatterplot3d)
+#!/usr/bin/env Rscript
+#
+# @package MiGA
+# @license Artistic-2.0
+#
 
-# Main function
-subclades <- function(ani_file, out_base, thr=1, ani=c()){
-   # Get ANI distances
-   cat("====", out_base, "\n")
-   if(missing(ani_file)){
-      a <- as.data.frame(ani)
-   } else {
-      a <- read.table(gzfile(ani_file), sep='\t', h=TRUE, as.is=T)
-   }
-   if(nrow(a)==0){
-      pdf(paste(out_base,'.pdf',sep=''), 7, 12)
-      plot(1,t='n',axes=F)
-      legend('center','No ANI data',bty='n')
-      dev.off()
-      file.create(paste(out_base,'.1.classif',sep=''))
-      file.create(paste(out_base,'.1.medoids',sep=''))
-      return(NULL)
-   }
-   ani.d <- enve.df2dist(cbind(a$a, a$b, 1-a$value/100), default.d=0.3)
-   ani.hc <- hclust(ani.d, method='ward.D2')
-   write.tree(as.phylo(ani.hc), 'miga-project.ani.nwk')
-   
-   # Silhouette
-   k <- 2:(length(labels(ani.d))-1)
-   s <- sapply(k, function(x) summary(silhouette(pam(ani.d, x)))$avg.width)
-   ds <- 10^(s[-c(1,length(s))]-(s[-length(s)+c(0,1)]+s[-c(1,2)])/2)
-   top.n <- head(k[order(c(-Inf,ds,-Inf), decreasing=T)],n=6)
+#= Load stuff
+argv <- commandArgs(trailingOnly=T)
+suppressPackageStartupMessages(library(ape))
+suppressPackageStartupMessages(library(vegan))
+suppressPackageStartupMessages(library(cluster))
+suppressPackageStartupMessages(library(parallel))
+suppressPackageStartupMessages(library(enveomics.R))
 
-   # Save "ANI-types"
-   ani.types <- c()
-   ani.medoids <- list()
-   for(i in 1:length(top.n)){
-      k_i <- top.n[i]
-      ani.cl <- pam(ani.d, k_i)
-      ani.types <- cbind(ani.types, ani.cl$clustering)
-      ani.medoids[[ i ]]  <- ani.cl$medoids
-   }
+#= Main function
+subclades <- function(ani_file, out_base, thr=1, ani=c()) {
+  say("==> Out base:", out_base, "<==")
+  
+  # Input arguments
+  if(missing(ani_file)){
+    a <- as.data.frame(ani)
+  }else{
+    a <- read.table(gzfile(ani_file), sep="\t", header=TRUE, as.is=TRUE)
+  }
+  if(nrow(a)==0){
+    generate_empty_files(out_base)
+    return(NULL)
+  }
+  
+  # Get ANI distances
+  say("Distances")
+  a$d <- 1-a$value/100
+  ani.d <- enve.df2dist(data.frame(a$a, a$b, a$d), default.d=max(a$d)*1.1)
+  ani.hc <- hclust(ani.d, method="ward.D2")
+  ani.ph <- as.phylo(ani.hc)
+  write.tree(as.phylo(ani.hc), paste(out_base, ".nwk", sep=""))
+  
+  # Silhouette
+  say("Silhouette")
+  k <- 2:min(length(labels(ani.d))-1, 100)
+  cl <- makeCluster(thr)
+  s <- parSapply(cl, k, function(x) {
+      library(cluster)
+      pam(ani.d, x, do.swap=FALSE, pamonce=1)$silinfo$avg.width
+    })
+  stopCluster(cl)
+  ds <- (s[-c(1,length(s))]-pmax(s[-length(s)+c(0,1)],s[-c(1,2)]))
+  top.n <- head(k[order(c(-Inf,ds,-Inf), decreasing=T)], n=1)
+  
+  # Classify genomes
+  say("Classify")
+  ani.cl <- pam(ani.d, top.n)
+  ani.types <- ani.cl$clustering
+  ani.medoids <- ani.cl$medoids
+  
+  # Generate graphic report
+  say("Graphic report")
+  pdf(paste(out_base, ".pdf", sep=""), 7, 12)
+  layout(1:4)
+  plot_distances(ani.d)
+  plot_silhouette(k, s, ds, top.n)
+  plot_clustering(ani.cl, ani.d, ani.types)
+  plot_tree(ani.ph, ani.types, ani.medoids)
+  dev.off()
 
-   # Generate graphic reports
-   pdf(paste(out_base,'.pdf',sep=''), 7, 12)
-   plotClusterAndMetadata(as.dendrogram(ani.hc), ani.types, main='ANI types')
-   ani.mds <- metaMDS(ani.d, k=3, autotransform=FALSE, parallel=thr, wascores=F)
-   layout(matrix(1:6, ncol=2))
-   for(i in 1:length(top.n)){
-      s <- scatterplot3d(ani.mds$points, pch=19, type='h',
-	 color=ggplotColours(top.n[i], alpha=1/2)[ani.types[,i]],
-	 cex.symbols=1/2, box=FALSE, lty.hplot=3,
-	 main=paste('NMDS of ANI distances with', top.n[i] ,'clusters'),
-	 angle=80, scale.y=3/2, las=2, xlab='', ylab='', zlab='')
-      for(cl in 1:top.n[i]){
-	 col <- ggplotColours(top.n[i])[cl]
-	 med <- s$xyz.convert(matrix(ani.mds$points[ ani.medoids[[i]][cl] , ],
-	    ncol=3))
-	 if(sum(ani.types[,i]==cl)>1){
-	    val <- s$xyz.convert(matrix(ani.mds$points[ ani.types[,i]==cl , ],
-	       ncol=3))
-	    arrows(x0=med$x, y0=med$y, x1=val$x, y1=val$y, length=0, col=col)
-	 }
-	 points(med, col=col, pch=19, cex=3/2)
-	 text(med, labels=cl, col='white', cex=2/3)
-      }
-   }
-   dev.off()
+  # Save results
+  say("Text report")
+  write.table(ani.medoids, paste(out_base, "medoids", sep="."),
+    quote=FALSE, col.names=FALSE, row.names=FALSE)
+  classif <- cbind(names(ani.types), ani.types, ani.medoids[ ani.types ], NA)
+  for(j in 1:nrow(classif)){
+    classif[j,4] <- 100 - as.matrix(ani.d)[classif[j,1], classif[j,3]]
+  }
+  write.table(classif, paste(out_base,"classif",sep="."),
+    quote=FALSE, col.names=FALSE, row.names=FALSE, sep="\t")
 
-   # Save results
-   for(i in 1:length(top.n)){
-      write.table(ani.medoids[[i]], paste(out_base,i,'medoids',sep='.'),
-	 quote=FALSE, col.names=FALSE, row.names=FALSE)
-      classif <- cbind(rownames(ani.types), ani.types[,i],
-	 ani.medoids[[i]][ ani.types[,i] ], NA)
-      for(j in 1:nrow(classif))
-	 classif[j,4] <- 100 - as.matrix(ani.d)[classif[j,1], classif[j,3]]
-      write.table(classif, paste(out_base,i,'classif',sep='.'),
-	 quote=FALSE, col.names=FALSE, row.names=FALSE, sep='\t')
-   }
+  # Recursive search
+  for(i in 1:top.n){
+    medoid <- ani.medoids[i]
+    ds_f <- names(ani.types)[ ani.types==i ]
+    say("Analyzing subclade", i, "with medoid:", medoid)
+    dir.create(paste(out_base, ".sc-", i, sep=""))
+    write.table(ds_f,
+      paste(out_base, ".sc-", i, "/miga-project.all",sep=""),
+      quote=FALSE, col.names=FALSE, row.names=FALSE)
+    if(length(ds_f) > 5){
+      a_f <- a[ (a$a %in% ds_f) & (a$b %in% ds_f), ]
+      subclades(out_base=paste(out_base, ".sc-", i, "/miga-project", sep=""),
+        thr=thr, ani=a_f)
+    }
+  }
+}
+
+#= Helper functions
+say <- function(...) { cat("[", date(), "]", ..., "\n") }
+
+generate_empty_files <- function(out_base) {
+  pdf(paste(out_base, ".pdf", sep=""), 7, 12)
+  plot(1, t="n", axes=F)
+  legend("center", "No data", bty="n")
+  dev.off()
+  file.create(paste(out_base,".1.classif",sep=""))
+  file.create(paste(out_base,".1.medoids",sep=""))
+}
+
+plot_silhouette <- function(k, s, ds, top.n) {
+  par(mar=c(4,5,1,5)+0.1)
+  plot(1, t="n", xlab="k (clusters)", ylab="", xlim=range(c(0,k)),
+    ylim=range(s), bty="n", xaxs="i", yaxt="n")
+  polygon(c(k[1], k, k[length(k)]), c(0,s,0), border=NA, col="grey80")
+  axis(2, fg="grey60", col.axis="grey60")
+  mtext("Average silhouette", side=2, line=3, col="grey60")
+  par(new=TRUE)
+  plot(1, t="n", xlab="", xaxt="n", ylab="", yaxt="n", xlim=range(c(0,k)),
+    ylim=range(ds), bty="n", xaxs="i")
+  points(k[-c(1,length(k))], ds, type="o", pch=16, col=rgb(1/2,0,0,3/4))
+  axis(4, fg="darkred", col.axis="darkred")
+  mtext("Silhouette gain", side=4, line=3, col="darkred")
+  abline(v=top.n, lty=2)
+}
+
+plot_distances <- function(dist) {
+  par(mar=c(5,4,1,2)+0.1)
+  hist(dist, border=NA, col="grey60", breaks=50, xlab="Distances", main="")
+}
 
-   # Explore subclades
-   for(i in 1:top.n[1]){
-      medoid <- ani.medoids[[1]][i]
-      ds_f <- rownames(ani.types)[ ani.types[,1]==i ]
-      cat("Analyzing subclade", i, "with medoid:", medoid, "\n")
-      cat("   ds_f: ", ds_f, "\n")
-      if(length(ds_f) > 5){
-	 a_f <- a[ (a$a %in% ds_f) & (a$b %in% ds_f), ]
-	 dir.create(paste(out_base,'.1.sc-',i,sep=''))
-	 write.table(ds_f,
-	    paste(out_base,'.1.sc-',i,'/miga-project.all',sep=''),
-	    quote=FALSE, col.names=FALSE, row.names=FALSE)
-	 cat("   looking for subclades within: ",
-	    out_base, ".1.sc-", i, "\n", sep="")
-	 subclades(
-	    out_base=paste(out_base,'.1.sc-',i,'/miga-project',sep=''),
-	    thr=thr, ani=a_f)
-      }
-   }
+plot_clustering <- function(cl, dist, types) {
+  par(mar=c(5,4,4,2)+0.1)
+  top.n <- length(cl$medoids)
+  col <- ggplotColours(top.n)
+  plot(silhouette(cl), col=col)
+  clusplot(cl, dist=dist, main="", col.p=col[types], lines=0)
 }
 
-# Ancillary functions
-plotClusterAndMetadata <- function(c,m,addLabels=TRUE,main='',type='factor'){
-   ps <- list()
-   ps[[1]] <- rectGrob(gp=gpar(col="white"))
-   if(length(type)==1) type <- rep(type, ncol(m))
-   if(addLabels){
-      m <- cbind(m, NA)
-      m[labels(c),ncol(m)] <- labels(c)
-      type[ncol(m)] <- 'label'
-   }
-   for(i in 1:ncol(m)){
-      df <- data.frame(lab=factor(labels(c),levels=labels(c)),
-	 feat=m[labels(c),i])
-      if(type[i]=='factor'){
-	 ps[[i+1]] <- ggplotGrob(ggplot(df,  aes(1, lab, fill=factor(feat))) +
-	    geom_tile() + geom_text(size=3/4, label=df$feat, x=.8) +
-	    scale_x_continuous(expand=c(0,0)) +
-	    theme(axis.title=element_blank(), panel.margin=unit(1,'points'),
-	       plot.margin=unit(c(40,-12,20,-12),'points'),
-	       axis.ticks=element_blank(), axis.text=element_blank(),
-	       legend.position="none"))
-      }else if(type[i]=='numeric'){
-	 ps[[i+1]] <- ggplotGrob(ggplot(df, aes(1,lab,fill=as.numeric(feat))) +
-	    geom_tile() + geom_text(size=3/4, label=df$feat, x=.8) +
-	    scale_x_continuous(expand=c(0,0)) +
-	    theme(axis.title=element_blank(), panel.margin=unit(1,'points'),
-	       plot.margin=unit(c(40,-12,20,-12),'points'),
-	       axis.ticks=element_blank(), axis.text=element_blank(),
-	       legend.position="none"))
-      }else if(type[i]=='label'){
-	 ps[[i+1]] <- ggplotGrob(ggplot(df,  aes(1, lab)) +
-	    geom_tile(fill='white') + geom_text(size=3/4, label=df$feat, x=.8) +
-	    theme(axis.title=element_blank(), panel.margin=unit(1,'points'),
-	       plot.margin=unit(c(40,-12,20,-12),'points'),
-	       axis.ticks=element_blank(), axis.text=element_blank(),
-	       legend.position="none"))
-      }else{
-	 stop('Unsupported type: ', type[i])
-      }
-   }
-   ps[[i+2]] <- ggplotGrob(ggplot(segment(dendro_data(c, type="rectangle"))) +
-      geom_segment(aes(x = x, y = y, xend = xend, yend = yend)) +
-      scale_x_continuous(expand=c(0,.5)) +
-      coord_flip() + theme_dendro() +
-         theme(axis.title=element_blank(), axis.ticks=element_blank(),
-	    plot.margin=unit(c(40,20,20,ifelse(addLabels,-35,-30)),'points'),
-            panel.margin=unit(0,'points'), axis.text=element_blank(),
-	    legend.position="none"))
-   maxHeights = do.call(grid::unit.pmax, lapply(ps, function(x) x$heights[2:5]))
-   for(g in ps) g$heights[2:5] <- as.list(maxHeights)
-   ps$nrow <- 1
-   ps$widths <- c(0.1,rep(.07,ncol(m)),1)
-   ps$main <- main
-   do.call(grid.arrange, ps)
-   return(ps)
+plot_tree <- function(phy, types, medoids){
+  layout(1)
+  top.n <- length(unique(types))
+  col <- ggplotColours(top.n)
+  is.medoid <- phy$tip.label %in% medoids
+  plot(phy, cex=ifelse(is.medoid, 1/3, 1/6),
+    font=ifelse(is.medoid, 2, 1),
+    tip.color=col[types[phy$tip.label]])
 }
 
 ggplotColours <- function(n=6, h=c(0, 360)+15, alpha=1){
-   if ((diff(h)%%360) < 1) h[2] <- h[2] - 360/n
-   hcl(h=seq(h[1], h[2], length=n), c=100, l=65, alpha=alpha)
+  if ((diff(h)%%360) < 1) h[2] <- h[2] - 360/n
+  hcl(h=seq(h[1], h[2], length=n), c=100, l=65, alpha=alpha)
 }
 
+#= Main
+subclades(ani_file=argv[1], out_base=argv[2],
+  thr=ifelse(is.na(argv[3]), 1, as.numeric(argv[3])))