From 5d43709a76b16f35a09a29675b20ae11503c8d54 Mon Sep 17 00:00:00 2001
From: Kelly Vincent <kpvincent@bx.psu.edu>
Date: Fri, 24 Jul 2009 17:06:12 -0400
Subject: [PATCH] Removed option to select indexing algorithm from BWA wrapper
 and have it automatically determine the algorithm by file size.

---
 tools/sr_mapping/bwa_wrapper.py  | 13 ++++++++++---
 tools/sr_mapping/bwa_wrapper.xml | 11 -----------
 2 files changed, 10 insertions(+), 14 deletions(-)

diff --git a/tools/sr_mapping/bwa_wrapper.py b/tools/sr_mapping/bwa_wrapper.py
index 3d3ab310f49..696555bc59c 100644
--- a/tools/sr_mapping/bwa_wrapper.py
+++ b/tools/sr_mapping/bwa_wrapper.py
@@ -17,7 +17,6 @@ def __main__():
     #Parse Command Line
     parser = optparse.OptionParser()
     parser.add_option('', '--ref', dest='ref', help='The reference genome to use or index')
-    parser.add_option('', '--indexingAlg', dest='indexingAlg', help='The algorithm to use while indexing')
     parser.add_option('', '--fastq', dest='fastq', help='The (forward) fastq file to use for the mapping')
     parser.add_option('', '--rfastq', dest='rfastq', help='The reverse fastq file to use for mapping if paired-end data')
     parser.add_option('', '--output', dest='output', help='The file to save the output (SAM format)')
@@ -53,10 +52,18 @@ def __main__():
             os.system('cp %s %s' % (options.ref, tmp_dir))
         except Exception, erf:
             stop_err('Error creating temp directory for indexing purposes\n' + str(erf))
+        try:
+            size = os.stat(options.ref).st_size
+            if size <= 2**30: 
+                indexingAlg = 'is'
+            else:
+                indexingAlg = 'bwtsw'
+        except:
+            indexingAlg = 'is'
         if options.fileType == 'solid':
-            indexing_cmds = '-c -a %s' % options.indexingAlg
+            indexing_cmds = '-c -a %s' % indexingAlg
         else:
-            indexing_cmds = '-a %s' % options.indexingAlg
+            indexing_cmds = '-a %s' % indexingAlg
         options.ref = os.path.join(tmp_dir,os.path.split(options.ref)[1])
         cmd1 = 'bwa index %s %s 2> /dev/null' % (indexing_cmds, options.ref)
         try:
diff --git a/tools/sr_mapping/bwa_wrapper.xml b/tools/sr_mapping/bwa_wrapper.xml
index ae4efaba4b6..2a28145bb35 100644
--- a/tools/sr_mapping/bwa_wrapper.xml
+++ b/tools/sr_mapping/bwa_wrapper.xml
@@ -4,10 +4,8 @@
     bwa_wrapper.py 
     #if $solidOrSolexa.solidRefGenomeSource.refGenomeSource == "history":
      --ref=$solidOrSolexa.solidRefGenomeSource.ownFile
-     --indexingAlg=$solidOrSolexa.solidRefGenomeSource.algorithm
     #else:
      --ref=$solidOrSolexa.solidRefGenomeSource.indices.value
-     --indexingAlg="None"
     #end if
     --fastq=$paired.input1
     #if $paired.sPaired == "paired":
@@ -74,10 +72,6 @@
 	      </param>
 	      <when value="history">
 	        <param name="ownFile" type="data" label="Select a reference genome" />
-	        <param name="algorithm" type="select" label="Select an indexing algorithm" help="IS works on databses 2GB or less, and is linear-time. BWT-SW works on database 10MB and larger, and trades speed for memory.">
-	          <option value="is">IS</option>
-	          <option value="bwtsw">BWT-SW</option>
-	        </param>
 	      </when>
 	      <when value="indexed">
  	    	<param name="indices" type="select" label="Select a reference genome">
@@ -98,10 +92,6 @@
 	      </param>
 	      <when value="history">
 	        <param name="ownFile" type="data" label="Select a reference genome" />
-	        <param name="algorithm" type="select" label="Select an indexing algorithm" help="IS works on databses 2GB or less, and is linear-time. BWT-SW works on database 10MB and larger, and trades speed for memory.">
-	          <option value="is">IS</option>
-	          <option value="bwtsw">BWT-SW</option>
-	        </param>
 	      </when>
 	      <when value="indexed">
  	        <param name="indices" type="select" label="Select a reference genome">
@@ -177,7 +167,6 @@
       <param name="solidSolexa" value="solid" />
       <param name="refGenomeSource" value="history" />
       <param name="ownFile" value="phiX.fa" />
-      <param name="algorithm" value="is" />
       <param name="sPaired" value="single" />
       <param name="input1" value="bwa_phiX_sanger.fastq" />
       <param name="source_select" value="pre_set" />