mostly script fixes and simple view tweaks

BIOFAB · Jun 5, 2012 · dea7c13 · dea7c13
1 parent 018eda1
commit dea7c13
Show file tree

Hide file tree

Showing 20 changed files with 703 additions and 248 deletions.
diff --git a/app/views/layouts/simple.html.erb b/app/views/layouts/simple.html.erb
@@ -96,8 +96,10 @@
 
     <div id='top_menu'>
       <ul>
+        <!--
         <li><%= link_to "Characterized bi-cistronic 5' UTRs", :action => "characterized_bds" %></li>
         <li><%= link_to "Characterized promoters", :action => "promoters" %></li>
+        -->
         <li><%= link_to "Randomized bi-cistronic 5' UTRs", :action => "randomized_bds" %></li>
         <li><%= link_to "Promoter vs. 5' UTR", :action => "index" %></li>
         <li class='last'><%= link_to "5' UTR vs. CDS", :action => "bd_gois" %></li>

diff --git a/app/views/simple/bd_gois.html.erb b/app/views/simple/bd_gois.html.erb
@@ -7,19 +7,19 @@
     <tr>
       <th>Show plasmid</th>
 
-      <th>Plasmid ID</th>
-      <th>Strain ID</th>
-      <th>5' UTR ID</th>
-      <th>CDS ID</th>
+      <th>Plasmid BIOFAB ID</th>
+      <th>Strain BIOFAB ID</th>
       <th>Strain performance</th>
       <th>Strain performance SD</th>
 
-      <th>5' UTR performance</th>
-      <th>5' UTR performance SD</th>
-      <th>5' UTR reliability</th>
+      <th>5' UTR BIOFAB ID</th>
+      <th>5' UTR name</th>
       <th>5' UTR sequence</th>
 
+      <th>CDS BIOFAB ID</th>
+      <th>CDS name</th>
       <th>CDS sequence</th>
+
     </tr>
   </thead>
   <tbody>
@@ -30,17 +30,17 @@
 
       <td><%= design.plasmid_biofab_id.gsub(/\.0$/, '') %></td>
       <td><%= design.strain_biofab_id.gsub(/\.0$/, '') %></td>
-      <td><%= design.fpu.biofab_id %></td>
-      <td><%= design.cds.biofab_id %></td>
-      <td><%= design.performance.round(2) %></td>
-      <td>?</td>
+      <td><%= design.fc_average.round(2) %></td>
+      <td><%= design.fc_sd.round(2) %></td>
+
+      <td><%= design.fpu_biofab_id %></td>
+      <td><%= design.fpu_name %></td>
+      <td><%= design.fpu_sequence %></td>
 
-      <td><%= design.fpu.part_performance.performance.round(2) %></td>
-      <td><%= design.fpu.part_performance.performance_sd.round(2) %></td>
-      <td><%= design.fpu.part_performance.reliability.round %></td>
-      <td><%= design.fpu.sequence %></td>
 
-      <td><%= design.cds.sequence %></td>
+      <td><%= design.cds_biofab_id %></td>
+      <td><%= design.cds_name %></td>
+      <td><%= design.cds_sequence %></td>
 
     </tr>
 

diff --git a/app/views/simple/characterized_bds.html.erb b/app/views/simple/characterized_bds.html.erb
@@ -3,7 +3,6 @@
     <tr>
       <th>5' UTR ID</th>
       <th>5' UTR performance</th>
-      <th>5' UTR performance SD</th>
       <th>5' UTR reliability</th>
       <th>5' UTR sequence</th>
     </tr>
@@ -17,7 +16,6 @@
       <td><%= fpu.biofab_id %></td>
 
       <td><%= fpu.part_performance.performance.round(2) %></td>
-      <td><%= fpu.part_performance.performance_sd.round(2) %></td>
       <td><%= fpu.part_performance.reliability.round %></td>
       <td><%= fpu.sequence %></td>
 

diff --git a/app/views/simple/index.html.erb b/app/views/simple/index.html.erb
@@ -72,21 +72,17 @@
     <tr>
       <th>Show plasmid</th>
 
-      <th>Plasmid ID</th>
-      <th>Strain ID</th>
-      <th>Promoter ID</th>
-      <th>5' UTR ID</th>
+      <th>Plasmid BIOFAB ID</th>
+      <th>Strain BIOFAB ID</th>
       <th>Strain performance</th>
       <th>Strain performance SD</th>
-
-      <th>Promoter performance</th>
-      <th>Promoter performance SD</th>
+      <th>Promoter BIOFAB ID</th>
+      <th>Promoter name</th>
       <th>Promoter sequence</th>
-
-      <th>5' UTR performance</th>
-      <th>5' UTR performance SD</th>
-      <th>5' UTR reliability</th>
+      <th>5' UTR BIOFAB ID</th>
+      <th>5' UTR name</th>
       <th>5' UTR sequence</th>
+
     </tr>
   </thead>
   <tbody>
@@ -98,18 +94,14 @@
 
       <td><%= design.plasmid_biofab_id.gsub(/\.0$/, '') %></td>
       <td><%= design.strain_biofab_id.gsub(/\.0$/, '') %></td>
-      <td><%= design.promoter.biofab_id %></td>
-      <td><%= design.fpu.biofab_id %></td>
-      <td><%= (design.performance_normalized * 100).round(2) %></td>
-      <td><%= (design.performance_sd_normalized * 100).round(2) %></td>
-      <td><%= design.promoter.part_performance.performance.round(2) %></td>
-      <td><%= design.promoter.part_performance.performance_sd.round(2) %></td>
-      <td><%= design.promoter.sequence %></td>
-
-      <td><%= design.fpu.part_performance.performance.round(2) %></td>
-      <td><%= design.fpu.part_performance.performance_sd.round(2) %></td>
-      <td><%= design.fpu.part_performance.reliability.round %></td>
-      <td><%= design.fpu.sequence %></td>
+      <td><%= design.performance.round(2)  %></td>
+      <td><%= design.performance_sd.round(2) %></td>
+      <td><%= design.promoter_biofab_id %></td>
+      <td><%= design.promoter_name %></td>
+      <td><%= design.promoter_sequence %></td>
+      <td><%= design.fpu_biofab_id %></td>
+      <td><%= design.fpu_name %></td>
+      <td><%= design.fpu_sequence %></td>
 
     </tr>
 

diff --git a/app/views/simple/promoters.html.erb b/app/views/simple/promoters.html.erb
@@ -3,7 +3,6 @@
     <tr>
       <th>Promoter ID</th>
       <th>Promoter performance</th>
-      <th>Promoter performance SD</th>
       <th>Promoter sequence</th>
     </tr>
   </thead>
@@ -15,7 +14,6 @@
       <td><%= promoter.biofab_id.gsub(/\.0$/, '') %></td>
 
       <td><%= promoter.part_performance.performance.round(2) %></td>
-      <td><%= promoter.part_performance.performance_sd.round(2) %></td>
       <td><%= promoter.sequence %></td>
 
     </tr>

diff --git a/app/views/simple/randomized_bds.html.erb b/app/views/simple/randomized_bds.html.erb
@@ -4,11 +4,11 @@
     <tr>
       <th>Show plasmid</th>
 
-      <th>Plasmid ID</th>
-      <th>Strain ID</th>
+      <th>Plasmid BIOFAB ID</th>
+      <th>Strain BIOFAB ID</th>
       <th>Strain performance</th>
       <th>Strain performance SD</th>
-      <th>5' UTR ID</th>
+      <th>5' UTR BIOFAB ID</th>
       <th>5' UTR sequence</th>
 
     </tr>
@@ -22,8 +22,8 @@
       <td><%= design.plasmid_biofab_id.gsub(/\.0$/, '') %></td>
       <td><%= design.strain_biofab_id.gsub(/\.0$/, '') %></td>
 
-      <td><%= design.fcs_normalized.round(2) %></td>
-      <td><%= design.fcs_sd_normalized.round(2) %></td>
+      <td><%= (design.fcs == 0) ? "no data" : design.fcs.round(2) %></td>
+      <td><%= (design.fcs_sd == 0) ? "no data" : design.fcs_sd.round(2) %></td>
 
       <td><%= design.biofab_id %></td>
       <td><%= design.sequence %></td>

diff --git a/db/migrate/20120528123005_bc_designs_add_performance_sd.rb b/db/migrate/20120528123005_bc_designs_add_performance_sd.rb
@@ -0,0 +1,8 @@
+class BcDesignsAddPerformanceSd < ActiveRecord::Migration
+  def up
+    add_column :bc_designs, :performance_sd, :float
+  end
+
+  def down
+  end
+end
diff --git a/db/migrate/20120602064928_design_add_fpu_name.rb b/db/migrate/20120602064928_design_add_fpu_name.rb
@@ -0,0 +1,8 @@
+class DesignAddFpuName < ActiveRecord::Migration
+  def up
+    add_column :designs, :fpu_name, :string
+  end
+
+  def down
+  end
+end
diff --git a/script/annotate_plasmids.rb b/script/annotate_plasmids.rb
@@ -21,7 +21,6 @@
 
 annots = Annotation.delete_all(["annotation_type_id = ?", anno_type.id])
 
-
 plasmid_type = PartType.find_by_name('Plasmid')
 raise "plasmid type not found" if !plasmid_type
 
@@ -48,8 +47,6 @@
 parts = Part.where(query_parm)
 raise "no parts found" if !parts || (parts.length == 0)
 
-
-
 plasmids.each do |plasmid| 
   next if plasmid.sequence.blank?
 
@@ -72,6 +69,8 @@
       annot_from = m.begin(0) + offset
       annot_to = m.end(0) + offset - 1
 
+      puts " -- annotating with: #{part.biofab_id} [#{part.part_type.name}]"
+
       anno = Annotation.new
       anno.parent_part = plasmid
       anno.part = part
@@ -82,7 +81,7 @@
 
       offset += m.end(0)
       mseq = mseq[offset..-1]
-
+      break if mseq.blank?
     end
   end
 
@@ -93,7 +92,24 @@
     annotations.each do |outer|
       next if outer == inner
       if (outer.from <= inner.from) && (outer.to >= inner.to)
-        keep = false
+        # if they have the same sequence, then make sure one and only one of them is added
+        if outer.part.sequence.upcase == inner.part.sequence.upcase
+          seq_in_final_list = false
+          final_annos.each do |final|
+            if final.part.sequence.upcase == inner.part.sequence.upcase
+              seq_in_final_list = true
+              break
+            end
+          end
+
+          if seq_in_final_list
+            keep = false
+#            puts " == dropping: #{inner.part.biofab_id} [#{inner.part.part_type.name}] because it is inside #{outer.part.biofab_id} [#{outer.part.part_type.name}]"
+          end
+        else
+#          puts " == dropping: #{inner.part.biofab_id} [#{inner.part.part_type.name}] because it is inside #{outer.part.biofab_id} [#{outer.part.part_type.name}]"
+          keep = false
+        end
       end
     end
     if keep

diff --git a/script/calculate_fpu_performance.rb b/script/calculate_fpu_performance.rb
@@ -1,43 +1,6 @@
 #!script/rails runner
-class Numeric
-  def square
-    self * self
-  end
-end
-
-module Enumerable
-
-  #  sum of an array of numbers
-  def sum
-    return self.inject(0){|acc,i|acc +i}
-  end
-
-  #  mean of an array of numbers
-  def mean
-    return self.sum/self.length.to_f
-  end
-
-  #  variance of an array of numbers
-  def variance2
-    avg=self.mean
-    sum=self.inject(0){|acc,i|acc +(i-avg)**2}
-    return(1/self.length.to_f*sum)
-  end
 
-  def squares
-    self.inject(0){|a,x|x.square+a}
-  end
-
-  def variance
-   self.squares.to_f/self.size - self.mean.square
-  end
-
-  #  standard deviation of an array of numbers
-  def standard_deviation
-    return Math.sqrt(self.variance)
-  end
-
-end  #  module Enumerable
+require 'script/math_stuff.rb'
 
 
 def find_obj(objs, fpu_part_id)

diff --git a/script/calculate_promoter_fpu_performance.rb b/script/calculate_promoter_fpu_performance.rb
@@ -14,11 +14,13 @@ def group_by_method(designs, method_name)
         :mean => nil,
         :sd => nil,
         :normalized_mean => nil,
-        :normalized_sd => nil
+        :normalized_sd => nil,
+        :strains => []
       }
     end
 
     objs[key][:numbers] << design.performance
+    objs[key][:strains] << design.strain_biofab_id
   end
 
   max = 0
@@ -31,8 +33,8 @@ def group_by_method(designs, method_name)
   end
 
   objs.each_key do |key|
-    objs[key][:normalized_mean] = objs[key][:mean] / max
-    objs[key][:normalized_sd] = objs[key][:sd] / max
+    objs[key][:normalized_mean] = objs[key][:mean]
+    objs[key][:normalized_sd] = objs[key][:sd]
   end
 
   objs
@@ -46,6 +48,8 @@ def group_by_method(designs, method_name)
 fpus = group_by_method(designs, :fpu_biofab_id)
 promoters = group_by_method(designs, :promoter_biofab_id)
 
+# TODO before merge, promoter-mean-center fpus and fpu-mean-center fpus
+
 objs = fpus.merge(promoters) do |key, val1, val2|
   raise "hash merge conflict for key #{key}"
 end
@@ -56,7 +60,7 @@ def group_by_method(designs, method_name)
 objs.each_key do |key|
   obj = objs[key]
 
-  puts "#{obj[:key]} - #{objs[key][:numbers].length} - #{obj[:normalized_mean] * 100}"
+  puts "#{obj[:key]} - #{objs[key][:numbers].length} - #{obj[:normalized_mean]} - #{obj[:normalized_sd]} - #{obj[:strains]}"
 
   part = Part.find_by_biofab_id(obj[:key])
   raise "Could not find part with biofab id #{obj[:key]}" if !part
@@ -66,8 +70,8 @@ def group_by_method(designs, method_name)
     perf.part_id = part.id
   end
 
-  perf.performance = obj[:normalized_mean] * 100
-  perf.performance_sd = obj[:normalized_sd] * 100
+  perf.performance = obj[:normalized_mean]
+  perf.performance_sd = obj[:normalized_sd]
 
   perf.save!
 end