From 422763cac75657d7dc01b7b174a04224dea7e124 Mon Sep 17 00:00:00 2001
From: Dave B <dave@bx.psu.edu>
Date: Fri, 8 Jun 2018 12:48:50 -0400
Subject: [PATCH] Update cummeRbund to retrieve genes and samples from input
 metadata.

---
 tools/cummerbund/cummeRbund.xml        | 27 ++++++++++++++------------
 tools/cummerbund/cummeRbund_macros.xml | 25 +++++++++++++++++++++---
 tools/cummerbund/cummeRbund_options.py | 11 -----------
 3 files changed, 37 insertions(+), 26 deletions(-)
 delete mode 100644 tools/cummerbund/cummeRbund_options.py

diff --git a/tools/cummerbund/cummeRbund.xml b/tools/cummerbund/cummeRbund.xml
index eae2d086..02cb4b3d 100644
--- a/tools/cummerbund/cummeRbund.xml
+++ b/tools/cummerbund/cummeRbund.xml
@@ -1,5 +1,5 @@
 <?xml version="1.0"?>
-<tool id="cummeRbund" name="cummeRbund" version="2.16.0">
+<tool id="cummeRbund" name="cummeRbund" version="2.16.0+galaxy1" profile="18.09">
     <description>visualize Cuffdiff output</description>
     <macros>
         <import>cummeRbund_macros.xml</import>
@@ -8,7 +8,6 @@
         <requirement type="package" version="1.0.1">r-argparse</requirement>
         <requirement type="package" version="2.16.0">bioconductor-cummerbund</requirement>
     </requirements>
-    <code file="cummeRbund_options.py"/>
     <command detect_errors="aggressive">
 <![CDATA[
 #for i, p in enumerate($plots):
@@ -22,7 +21,8 @@
         #if $p.plot.multiple_genes.multiple_genes_selector == "yes":
             --features $p.plot.multiple_genes.features --gene_selector
             #for gene in $p.plot.multiple_genes.genes:
-                --genes ${gene.gene_id}
+                #set $gene_id = str($gene.gene_id).split(': ')[0]
+                --genes $gene_id
             #end for
         #end if
     #elif $p.plot.type == "maplot":
@@ -32,7 +32,8 @@
         #if $p.plot.multiple_genes.multiple_genes_selector == "yes":
             --features $p.plot.multiple_genes.features --gene_selector
             #for gene in $p.plot.multiple_genes.genes:
-                --genes ${gene.gene_id}
+                #set $gene_id = str($gene.gene_id).split(': ')[0]
+                --genes $gene_id
             #end for
         #end if
     #elif $p.plot.type == "heatmap":
@@ -40,21 +41,24 @@
         #if len($p.plot.genes) > 0:
             --gene_selector
             #for gene in $p.plot.genes:
-                --genes ${gene.gene_id}
+                #set $gene_id = str($gene.gene_id).split(': ')[0]
+                --genes $gene_id
             #end for
         #end if
     #elif $p.plot.type == "cluster":
         --features $p.plot.features --k $p.plot.k --iter_max $p.plot.iter_max
         #if len($p.plot.genes) > 0:
             #for gene in $p.plot.genes:
-                --genes ${gene.gene_id}
+                #set $gene_id = str($gene.gene_id).split(': ')[0]
+                --genes $gene_id
             #end for
         #end if
     #elif $p.plot.type in [ "expressionplot", "expressionbarplot" ]:
         #if $p.plot.type == "expressionplot":
             $p.plot.draw_summary
         #end if
-        --features $p.plot.features $p.plot.error_bars --genes ${p.plot.gene_id} $p.plot.replicates $p.plot.log10
+        #set $gene_id = str($p.plot.gene_id).split(': ')[0]
+        --features $p.plot.features $p.plot.error_bars --genes $gene_id $p.plot.replicates $p.plot.log10
     #end if
     #if $p.plot.type == "density":
         $p.plot.log10
@@ -63,7 +67,7 @@
 #end for
 ]]></command>
     <inputs>
-        <param name="input_database" type="data" format="sqlite" label="Select backend database (sqlite)" />
+        <param name="input_database" type="data" format="cuffdiff.sqlite" label="Select backend database (sqlite)" />
         <repeat name="plots" title="Plots">
             <param name="width" type="integer" value="1280" label="The width of the image"/>
             <param name="height" type="integer" value="960" label="The height of the image"/>
@@ -109,8 +113,7 @@
                     <expand macro="multiple_genes_conditional" />
                 </when>
                 <when value="volcano">
-                    <param name="x" type="select" label="First sample name for comparison" dynamic_options="get_samples(input_database.dataset.file_name)" />
-                    <param name="y" type="select" label="Second sample name for comparison" dynamic_options="get_samples(input_database.dataset.file_name)" />
+                    <expand macro="xy_selector" />
                     <expand macro="multiple_genes_conditional" />
                 </when>
                 <when value="heatmap">
@@ -142,7 +145,7 @@
                 <when value="scatterMatrix" />
                 <when value="expressionplot">
                     <expand macro="features_selector" />
-                    <param name="gene_id" type="select" label="Gene ID" dynamic_options="get_genes(input_database.dataset.file_name)" />
+                    <expand macro="gene_selector" />
                     <param name="draw_summary" type="boolean" truevalue="--summary" falsevalue="" checked="False" label="Draw a 'summary' line with mean FPKM values for each condition?"/>
                     <param name="error_bars" type="boolean" truevalue="--error_bars" falsevalue="" checked="True" label="Draw error bars?"/>
                     <expand macro="replicates_checkbox" />
@@ -150,7 +153,7 @@
                 </when>
                 <when value="expressionbarplot">
                     <expand macro="features_selector" />
-                    <param name="gene_id" type="select" label="Gene ID" dynamic_options="get_genes(input_database.dataset.file_name)" />
+                    <expand macro="gene_selector" />
                     <param name="error_bars" type="boolean" truevalue="--error_bars" falsevalue="" checked="True" label="Draw error bars?"/>
                     <expand macro="replicates_checkbox" />
                     <expand macro="log10_checkbox" />
diff --git a/tools/cummerbund/cummeRbund_macros.xml b/tools/cummerbund/cummeRbund_macros.xml
index aee71742..952ce8db 100644
--- a/tools/cummerbund/cummeRbund_macros.xml
+++ b/tools/cummerbund/cummeRbund_macros.xml
@@ -6,14 +6,33 @@
     <macro name="log10_checkbox">
         <param name="log10" type="boolean" truevalue="--log10" falsevalue="" checked="True" label="Apply log10 transformation on FPKM values?"/>
     </macro>
+    <macro name="samples_meta">
+        <options>
+            <filter type="data_meta" ref="input_database" key="samples" />
+        </options>
+    </macro>
+    <macro name="genes_meta">
+        <options>
+            <filter type="data_meta" ref="input_database" key="genes" />
+        </options>
+    </macro>
+    <macro name="gene_selector">
+        <param name="gene_id" type="select" label="Gene ID">
+            <expand macro="genes_meta" />
+        </param>
+    </macro>
     <macro name="xy_selector">
-        <param name="x" type="select" label="Sample name for x axis" dynamic_options="get_samples(input_database.dataset.file_name)" />
-        <param name="y" type="select" label="Sample name for y axis" dynamic_options="get_samples(input_database.dataset.file_name)" />
+        <param name="x" type="select" label="Sample name for x axis">
+            <expand macro="samples_meta" />
+        </param>
+        <param name="y" type="select" label="Sample name for y axis">
+            <expand macro="samples_meta" />
+        </param>
     </macro>
     <macro name="genes_selector">
         <repeat name="genes" title="Genes" min="2">
             <!-- Cannot create a heatmap for less than two genes -->
-            <param name="gene_id" type="select" label="Gene ID" dynamic_options="get_genes(input_database.dataset.file_name)" />
+            <expand macro="gene_selector" />
         </repeat>
     </macro>
     <macro name="features_selector">
diff --git a/tools/cummerbund/cummeRbund_options.py b/tools/cummerbund/cummeRbund_options.py
deleted file mode 100644
index f2e214bd..00000000
--- a/tools/cummerbund/cummeRbund_options.py
+++ /dev/null
@@ -1,11 +0,0 @@
-import sqlite3
-
-def get_genes( database_path ):
-    conn = sqlite3.connect( database_path )
-    gene_ids = conn.execute( 'SELECT gene_short_name, gene_id FROM genes ORDER BY gene_short_name' )
-    return [ ( gene_id[ 0 ], gene_id[ 1 ], False ) for gene_id in gene_ids ]
-
-def get_samples( database_path ):
-    conn = sqlite3.connect( database_path )
-    samples = conn.execute( 'SELECT sample_name FROM samples ORDER BY sample_name' )
-    return [ ( sample[ 0 ], sample[ 0 ], False ) for sample in samples ]