Merge pull request RECETOX#533 from zargham-ahmad/issue508

Update WaveICA to outputs
zargham-ahmad · May 7, 2024 · 84de3bc · 84de3bc
2 parents 722ce0a + c039521
commit 84de3bc
Show file tree

Hide file tree

Showing 4 changed files with 17 additions and 21 deletions.
diff --git a/tools/waveica/macros.xml b/tools/waveica/macros.xml
@@ -99,7 +99,11 @@
 
     <xml name="outputs">
         <outputs>
-             <data format_source="data" name="normalized_data"/>
+            <data name="normalized_data" format="tsv">
+                <change_format>
+                    <when input_dataset="data" attribute="ext" value="parquet" format="parquet" />
+                </change_format>
+            </data>
         </outputs>
     </xml>
 

diff --git a/tools/waveica/test-data/normalized_data.csv b/tools/waveica/test-data/normalized_data.csv
diff --git a/tools/waveica/waveica.xml b/tools/waveica/waveica.xml
@@ -1,4 +1,4 @@
-<tool id="waveica" name="WaveICA" version="@TOOL_VERSION@+galaxy4" python_template_version="3.5">
+<tool id="waveica" name="WaveICA" version="@TOOL_VERSION@+galaxy5" profile="21.09">
     <description>removal of batch effects for untargeted metabolomics data</description>
     <macros>
         <import>macros.xml</import>
@@ -103,7 +103,7 @@
             <param name="t" value="0.05"/>
             <param name="t2" value="0.05"/>
             <param name="alpha" value="0"/>
-            <output name="normalized_data" file="normalized_data.csv"/>
+            <output name="normalized_data" file="normalized_data.tsv" ftype="tsv"/>
         </test>
         <test><!-- TEST 2 -->
             <param name="data" value="input_data.tsv" ftype="tsv"/>
@@ -114,7 +114,7 @@
             <param name="t" value="0.05"/>
             <param name="t2" value="0.05"/>
             <param name="alpha" value="0"/>
-            <output name="normalized_data" file="normalized_data.tsv"/>
+            <output name="normalized_data" file="normalized_data.tsv" ftype="tsv"/>
         </test>
         <test><!-- TEST 3 -->
             <param name="data" value="input_data.parquet" ftype="parquet"/>
@@ -125,7 +125,7 @@
             <param name="t" value="0.05"/>
             <param name="t2" value="0.05"/>
             <param name="alpha" value="0"/>
-            <output name="normalized_data" file="normalized_data.parquet"/>
+            <output name="normalized_data" file="normalized_data.parquet" ftype="parquet"/>
         </test>
         <test><!-- TEST 4 -->
             <param name="input_choice" value="2"/>
@@ -138,7 +138,7 @@
             <param name="t" value="0.05"/>
             <param name="t2" value="0.05"/>
             <param name="alpha" value="0"/>
-            <output name="normalized_data" file="normalized_data.csv"/>
+            <output name="normalized_data" file="normalized_data.tsv" ftype="tsv"/>
         </test>
         <test><!-- TEST 5 -->
             <param name="input_choice" value="2"/>
@@ -151,7 +151,7 @@
             <param name="t" value="0.05"/>
             <param name="t2" value="0.05"/>
             <param name="alpha" value="0"/>
-            <output name="normalized_data" file="normalized_data.tsv"/>
+            <output name="normalized_data" file="normalized_data.tsv" ftype="tsv"/>
         </test>
         <test><!-- TEST 6 -->
             <param name="input_choice" value="2"/>
@@ -164,7 +164,7 @@
             <param name="t" value="0.05"/>
             <param name="t2" value="0.05"/>
             <param name="alpha" value="0"/>
-            <output name="normalized_data" file="normalized_data.parquet" compare="sim_size" delta="200"/>
+            <output name="normalized_data" file="normalized_data.parquet" compare="sim_size" delta="200" ftype="parquet"/>
         </test>
         <test><!-- TEST 7 -->
             <param name="input_choice" value="2"/>
@@ -178,7 +178,7 @@
             <param name="t" value="0.05"/>
             <param name="t2" value="0.05"/>
             <param name="alpha" value="0"/>
-            <output name="normalized_data" file="normalized_data.parquet" compare="sim_size" delta="200"/>
+            <output name="normalized_data" file="normalized_data.parquet" compare="sim_size" delta="200" ftype="parquet"/>
         </test>
         <test><!-- TEST 8 -->
             <param name="input_choice" value="2"/>
@@ -192,7 +192,7 @@
             <param name="t" value="0.05"/>
             <param name="t2" value="0.05"/>
             <param name="alpha" value="0"/>
-            <output name="normalized_data" file="normalized_data.csv"/>
+            <output name="normalized_data" file="normalized_data.tsv" ftype="tsv"/>
         </test>
         <!-- The following test has different results on three platform I've tried -->
         <!-- <test>

diff --git a/tools/waveica/waveica_wrapper.R b/tools/waveica/waveica_wrapper.R
@@ -231,15 +231,13 @@ exclude_group <- function(data, group) {
 }
 
 store_data <- function(data, output, ext) {
-  if (ext == "csv") {
-    write.csv(data, file = output, row.names = FALSE, quote = FALSE)
-  } else if (ext == "tsv") {
+  if (ext == "parquet") {
+    arrow::write_parquet(data, output)
+  } else {
     write.table(data,
       file = output, sep = "\t",
       row.names = FALSE, quote = FALSE
     )
-  } else {
-    arrow::write_parquet(data, sink = output)
   }
   cat("Normalization has been completed.\n")
 }