From 39f811eff7f29b42b397d44fa24fbc8b5f69d610 Mon Sep 17 00:00:00 2001
From: Greg Von Kuster <greg@bx.psu.edu>
Date: Mon, 7 May 2007 18:59:15 +0000
Subject: [PATCH] Added new Subtract Whole Query tool.  Also some miscellaneous
 code clean-up.

---
 test-data/1.txt                         | 10 +++
 tool_conf.xml.sample                    |  3 +-
 tools/data_source/upload_code.py        | 11 ----
 tools/filters/compare.xml               |  2 +-
 tools/new_operations/subtract_query.py  | 45 ++++++++++++++
 tools/new_operations/subtract_query.xml | 82 +++++++++++++++++++++++++
 6 files changed, 140 insertions(+), 13 deletions(-)
 create mode 100644 test-data/1.txt
 delete mode 100644 tools/data_source/upload_code.py
 create mode 100644 tools/new_operations/subtract_query.py
 create mode 100644 tools/new_operations/subtract_query.xml

diff --git a/test-data/1.txt b/test-data/1.txt
new file mode 100644
index 00000000000..9e8397bcbd6
--- /dev/null
+++ b/test-data/1.txt
@@ -0,0 +1,10 @@
+chr1    4225    19670
+chr10   6       8
+chr1    24417   24420
+chr6_hla_hap2   0       150
+chr2    1       5
+chr10   2       10
+chr1    30      55
+chrY    1       20
+chr1    1225979 42287290
+chr10   7       8
diff --git a/tool_conf.xml.sample b/tool_conf.xml.sample
index 9b5b8e886d8..ad5f815b8dc 100644
--- a/tool_conf.xml.sample
+++ b/tool_conf.xml.sample
@@ -42,12 +42,13 @@
     <tool file="filters/headWrapper.xml" />
     <tool file="filters/tailWrapper.xml" />
   </section>
-  <section name="Filter, Sort, Join and Compare" id="filter">
+  <section name="Filter, Sort, Join, Compare, Subtract" id="filter">
     <tool file="stats/filtering.xml" />
     <tool file="filters/sorter.xml" />
     <tool file="filters/grep.xml" />
     <tool file="filters/joiner.xml" />
     <tool file="filters/compare.xml"/>
+    <tool file="new_operations/subtract_query.xml"/>
   </section>
   <section name="Convert Formats" id="convert">
     <tool file="filters/maf/maf_to_fasta_multiple_sets.xml" />
diff --git a/tools/data_source/upload_code.py b/tools/data_source/upload_code.py
deleted file mode 100644
index 0bdf2e3b0af..00000000000
--- a/tools/data_source/upload_code.py
+++ /dev/null
@@ -1,11 +0,0 @@
-#Provides Upload tool with access to list of available builds
-import galaxy.util
-builds = []
-
-#Read build names and keys from galaxy.util
-for dbkey, build_name in galaxy.util.dbnames:
-    builds.append((build_name,dbkey,False))
-
-#Return available builds
-def get_available_builds():
-    return builds
diff --git a/tools/filters/compare.xml b/tools/filters/compare.xml
index b6eda4809cd..ec87a8aa724 100644
--- a/tools/filters/compare.xml
+++ b/tools/filters/compare.xml
@@ -65,7 +65,7 @@ Finding lines of the **First query** whose 4th column matching the 1st column of
   chr1 10 20 geneA 
   chr1 50 80 geneB
 
-Converesely, using options **Non Matching rows of First query** on same fields will yield::
+Conversely, using options **Non Matching rows of First query** on same fields will yield::
 
   chr5 10 40 geneL
 
diff --git a/tools/new_operations/subtract_query.py b/tools/new_operations/subtract_query.py
new file mode 100644
index 00000000000..ff7fa458978
--- /dev/null
+++ b/tools/new_operations/subtract_query.py
@@ -0,0 +1,45 @@
+#!/usr/bin/env python2.4
+# Greg Von Kuster
+
+"""
+Subtract an entire query from another query
+usage: %prog in_file_1 in_file_2 out_file
+"""
+
+import sys, sets
+import cookbook.doc_optparse
+
+
+def get_lines(fname):
+    lines = set([])
+    for i, line in enumerate(file(fname)):
+        line = line.strip()
+        lines.add( line )
+    return lines
+
+def main():   
+    # Parsing Command Line here
+    options, args = cookbook.doc_optparse.parse( __doc__ )
+    
+    try:
+        inp1_file, inp2_file, out_file = args
+    except:
+        cookbook.doc_optparse.exception()
+
+    try:
+        fo = open(out_file,'w')
+    except:
+        print >> sys.stderr, "Unable to open output file"
+        sys.exit()
+
+    lines1 = get_lines(inp1_file)
+    lines2 = get_lines(inp2_file)
+    lines1.difference_update(lines2)
+    
+    for line in lines1:
+        print >> fo, line
+
+    fo.close()
+    
+if __name__ == "__main__":
+    main()
diff --git a/tools/new_operations/subtract_query.xml b/tools/new_operations/subtract_query.xml
new file mode 100644
index 00000000000..758785613d9
--- /dev/null
+++ b/tools/new_operations/subtract_query.xml
@@ -0,0 +1,82 @@
+<tool id="subtract_query1" name="Subtract Whole Query">
+  <description>from another query</description>
+  <command interpreter="python2.4">subtract_query.py $input1 $input2 $output</command>
+  <inputs>
+    <param format="text" name="input2" type="data" help="Second query">
+      <label>Subtract</label>
+    </param>
+    
+    <param format="text" name="input1" type="data" help="First query">
+      <label>from</label>
+    </param>
+   </inputs>
+  <outputs>
+    <data format="input" name="output" metadata_source="input1" />
+  </outputs>
+  <tests>
+    <test>
+      <param name="input1" value="1.txt" />
+      <param name="input2" value="2.txt" />
+      <output name="output" file="subtract-query-1.dat" />
+    </test>
+    <test>
+      <param name="input1" value="eq-showbeginning.dat" />
+      <param name="input2" value="eq-showtail.dat" />
+      <output name="output" file="subtract-query-2.dat" />
+    </test>
+    <test>
+      <param name="input1" value="eq-showbeginning.dat" />
+      <param name="input2" value="eq-removebeginning.dat" />
+      <output name="output" file="subtract-query-3.dat" />
+    </test>
+  </tests>
+  <help>
+
+.. class:: infomark
+
+**TIP:** This tool complements the tool in the **Operate on Genomic Intervals** tool set which subtracts the intervals of two queries.
+
+
+-----
+
+**Syntax**
+
+This tool subtracts an entire query from another query.  Any text format is valid.
+
+-----
+
+**Example**
+
+If this is the **First query**::
+
+  chr1    4225    19670
+  chr10   6       8
+  chr1    24417   24420
+  chr6_hla_hap2   0       150
+  chr2    1       5
+  chr10   2       10
+  chr1    30      55
+  chrY    1       20
+  chr1    1225979 42287290
+  chr10   7       8 
+
+and this is the **Second query**::
+
+  chr1    4225    19670
+  chr10   6       8
+  chr1    24417   24420
+  chr6_hla_hap2   0       150
+  chr2    1       5
+  chr1    30      55
+  chrY    1       20
+  chr1    1225979 42287290
+
+Subtracting the **Second query** from the **First query** will yield::
+
+  chr10   7       8 
+  chr10   2       10
+
+Conversely, subtracting the **First query** from the **Second query** will result in an empty dataset.
+
+</help>
+</tool>
\ No newline at end of file