Rewrite random lines tool. Does two passes through input file: first pass determines line offsets/count, offsets are randomly removed, second pass writes out lines by seeking to remaining offsets. The random seed can now be optionally specified.

tools/filters/random_lines_two_pass.py

@@ -0,0 +1,51 @@
+#!/usr/bin/env python
+#Dan Blankenberg
+#Selects N random lines from a file and outputs to another file, maintaining original line order
+#allows specifying a seed
+#does two passes to determine line counts and offsets, and then to output contents
+
+import optparse, random 
+
+def __main__():
+    #Parse Command Line
+    parser = optparse.OptionParser()
+    parser.add_option( '-s', '--seed', dest='seed', action='store', type="string", default=None, help='Set the random seed.' )
+    (options, args) = parser.parse_args()
+    
+    assert len( args ) == 3, "Invalid command line specified."
+    
+    input = open( args[0] )
+    output = open( args[1], 'wb' )
+    num_lines = int( args[2] )
+    assert num_lines > 0, "You must select at least one line."
+    
+    if options.seed is not None:
+        random.seed( options.seed )
+    
+    #get line offsets
+    line_offsets = []
+    while True:
+        offset = input.tell()
+        if input.readline():
+            line_offsets.append( offset )
+        else:
+            break
+    
+    total_lines = len( line_offsets )
+    assert num_lines <= total_lines, "Error: asked to select more lines (%i) than there were in the file (%i)." % ( num_lines, total_lines )
+    
+    #get random line offsets
+    while len( line_offsets ) > num_lines:
+        line_offsets.pop( random.randint( 0, len( line_offsets ) - 1 ) )
+    
+    #write out random lines
+    for line_offset in line_offsets:
+        input.seek( line_offset )
+        output.write( input.readline() )
+    input.close()
+    output.close()
+    print "Kept %i of %i total lines." % ( num_lines, total_lines )
+    if options.seed is not None:
+        print 'Used random seed of "%s".' % options.seed
+    
+if __name__=="__main__": __main__()

tools/filters/randomlines.xml

@@ -1,9 +1,25 @@
-<tool id="random_lines1" name="Select random lines">
+<tool id="random_lines1" name="Select random lines" version="2.0.0">
   <description>from a file</description>
-  <command interpreter="python">randomlines.py $input $num_lines $out_file1</command>
+  <command interpreter="python">random_lines_two_pass.py "${input}" "${out_file1}" "${num_lines}"
+  #if str( $seed_source.seed_source_selector ) == "set_seed":
+      --seed "${seed_source.seed}"
+  #end if
+  </command>
   <inputs>
     <param name="num_lines" size="5" type="integer" value="1" label="Randomly select" help="lines"/>
     <param format="txt" name="input" type="data" label="from"/>
+    <conditional name="seed_source">
+      <param name="seed_source_selector" type="select" label="Set a random seed">
+        <option value="no_seed" selected="True">Don't set seed</option>
+        <option value="set_seed">Set seed</option>
+      </param>
+      <when value="no_seed">
+        <!-- Do nothing here -->
+      </when>
+      <when value="set_seed"> 
+        <param name="seed" type="text" label="Random seed" />
+      </when>
+    </conditional>
   </inputs>
   <outputs>
     <data format="input" name="out_file1" metadata_source="input"/>
@@ -12,8 +28,16 @@
     <test>
       <param name="num_lines" value="65"/>
       <param name="input" value="1.bed"/>
+      <param name="seed_source_selector" value="no_seed"/>
       <output name="out_file1" file="1.bed"/>
     </test>
+    <test>
+      <param name="num_lines" value="1"/>
+      <param name="input" value="1.bed"/>
+      <param name="seed_source_selector" value="set_seed"/>
+      <param name="seed" value="asdf"/>
+      <output name="out_file1" file="1_bed_random_lines_1_seed_asdf_out.bed"/>
+    </test>
   </tests>
   <help>