Mzmine 3.9 (galaxyproject#5622)

* mzmine: bump to 3.9 * mzmine: limit number of cores fro JVM * mzmine: add possibility to use local CSV database * mzmine: reduce amount of remote test data and version the remaining * mzmine: comment tests don't work unless planemo can use the 23.1 galaxy packages * Change quoting Co-authored-by: Björn Grüning <[email protected]> * improve help message * allow also tsv and txt technically all this is allowed, batch file can specify the delimiter the example uses ; --------- Co-authored-by: Björn Grüning <[email protected]>
pavanvidem · Nov 10, 2023 · 889d26f · 889d26f
1 parent 3a3dfd6
commit 889d26f
Show file tree

Hide file tree

Showing 5 changed files with 9,199 additions and 10 deletions.
diff --git a/tools/mzmine/mzmine_batch.xml b/tools/mzmine/mzmine_batch.xml
@@ -1,13 +1,14 @@
 <tool id="mzmine_batch" name="MZMine batch" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="21.05">
     <macros>
-        <token name="@TOOL_VERSION@">3.6.0</token>
+        <token name="@TOOL_VERSION@">3.9.0</token>
         <token name="@VERSION_SUFFIX@">0</token>
     </macros>
     <xrefs>
         <xref type="bio.tools">mzmine</xref>
     </xrefs>
     <requirements>
         <requirement type="package" version="@TOOL_VERSION@">mzmine</requirement>
+        <requirement type="package" version="3.11">python</requirement>
     </requirements>
     <command detect_errors="exit_code"><![CDATA[
     #import re
@@ -30,9 +31,16 @@
     ## modify output paths in batch XML to relative path
     ## - any output will be redirected to ./output/BASENAME
     mkdir -p output/ &&
-    
-    sed -e 's@<current_file>.*[\\\/]\([^\\\/]\+\)</current_file>@<current_file>./output/\1</current_file>@' '$batch' > batch.xml &&
+    python '$__tool_directory__/set_path.py'
+        --input '$batch'
+        --output batch.xml
+    #if $localdb
+        --localdb '$localdb'
+    #end if
+    &&
 
+    ## https://github.com/mzmine/mzmine3/issues/1534
+    JAVA_OPTS="-XX:ActiveProcessorCount=\${GALAXY_SLOTS:-1}"
     mzmine
         --batch 'batch.xml'
         --input 'input.txt'
@@ -54,20 +62,18 @@
         <param argument="--batch" type="data" format="xml" label="MZMine batch file" help="XML batch file (ideally created with version @TOOL_VERSION@)"/>
         <param argument="--input" type="data" format="mzml,mzxml,csv" multiple="true" label="Spectra (mzml)"/>
         <param argument="--libraries" type="data" format="json,mgf,msp" multiple="true" optional="true" label="Spectral libraries"/>
+        <param name="localdb" type="data" format="csv,tsv,tabular,txt" optional="true" label="Local database" help="For 'Custom database search'"/>
     </inputs>
     <outputs>
         <collection name="output" type="list">
             <discover_datasets pattern="__name_and_ext__" directory="output"/>
         </collection>
     </outputs>
     <tests>
-        <test>
+        <!-- <test>
             <param name="batch" value="test_small.xml" ftype="xml"/>
-            <param name="input" value="DOM_a.mzML,DOM_b.mzXML"/>
+            <param name="input" location="https://raw.githubusercontent.com/mzmine/mzmine3/v@TOOL_VERSION@/src/test/resources/rawdatafiles/DOM_a.mzML,https://raw.githubusercontent.com/mzmine/mzmine3/v@TOOL_VERSION@/src/test/resources/rawdatafiles/DOM_b.mzXML"/>
             <param name="libraries" value="GNPS-FAULKNERLEGACY.json" ftype="json"/>
-            <!-- <param name="input" location="https://raw.githubusercontent.com/mzmine/mzmine3/master/src/test/resources/rawdatafiles/DOM_a.mzML"/> -->
-            <!-- <param name="input" location="https://raw.githubusercontent.com/mzmine/mzmine3/master/src/test/resources/rawdatafiles/DOM_a.mzML,https://raw.githubusercontent.com/mzmine/mzmine3/master/src/test/resources/rawdatafiles/DOM_b.mzXML"/>
-            <param name="libraries" location="https://external.gnps2.org/gnpslibrary/GNPS-FAULKNERLEGACY.json" ftype="json"/> -->
             <output_collection name="output" count="3" type="list">
                 <element name="test2_iimn_gnps" ftype="mgf">
                     <assert_contents>
@@ -88,7 +94,51 @@
                     </assert_contents>
                 </element>
             </output_collection>
-        </test>
+            <assert_stderr>
+                <has_text text="INFO: Starting processing of task Local CSV identification using database" negate="true"/>
+                <has_text text="INFO: Finished a batch of 11 steps"/>
+            </assert_stderr>
+        </test> -->
+        <!-- test with localdb -->
+        <!-- <test>
+            <param name="batch" value="test_localdatabase.xml" ftype="xml"/>
+            <param name="input" location="https://raw.githubusercontent.com/mzmine/mzmine3/v@TOOL_VERSION@/src/test/resources/rawdatafiles/DOM_a.mzML,https://raw.githubusercontent.com/mzmine/mzmine3/v@TOOL_VERSION@/src/test/resources/rawdatafiles/DOM_b.mzXML"/>
+            <param name="libraries" value="GNPS-FAULKNERLEGACY.json" ftype="json"/>
+            <param name="localdb" value="local_database.csv" ftype="csv"/>
+            <output_collection name="output" count="3" type="list">
+                <element name="test2_iimn_gnps" ftype="mgf">
+                    <assert_contents>
+                        <has_n_lines n="2103"/>
+                        <has_text text="BEGIN IONS"/>
+                    </assert_contents>
+                </element>
+                <element name="test2_iimn_gnps_quant" ftype="csv">
+                    <assert_contents>
+                        <has_n_lines n="34"/>
+                        <has_n_columns sep="," n="16"/>
+                    </assert_contents>
+                </element>
+                <element name="test2_sirius" ftype="mgf">
+                    <assert_contents>
+                        <has_n_lines n="54406"/>
+                        <has_text text="BEGIN IONS"/>
+                    </assert_contents>
+                </element>
+            </output_collection>
+            <assert_stderr>
+                <has_text text="INFO: Starting processing of task Local CSV identification using database"/>
+                <has_text text="INFO: Finished a batch of 12 steps"/>
+            </assert_stderr>
+        </test> -->
+        <!-- test with localdb in batch, but user does not provide one-->
+        <!-- <test expect_failure="true" expect_exit_code="1">
+            <param name="batch" value="test_localdatabase.xml" ftype="xml"/>
+            <param name="input" location="https://raw.githubusercontent.com/mzmine/mzmine3/v@TOOL_VERSION@/src/test/resources/rawdatafiles/DOM_a.mzML,https://raw.githubusercontent.com/mzmine/mzmine3/v@TOOL_VERSION@/src/test/resources/rawdatafiles/DOM_b.mzXML"/>
+            <param name="libraries" value="GNPS-FAULKNERLEGACY.json" ftype="json"/>
+            <assert_stderr>
+                <has_text text="Batch file contains LocalCSVDatabaseSearchModule but no local DB CSV file given"/>
+            </assert_stderr>
+        </test> -->
     </tests>
     <help><![CDATA[
         MZmine 3 is an open-source software for mass-spectrometry data processing, with the main focus on LC-MS data.
@@ -102,6 +152,13 @@
         - Input spectra in mzml format
         - Spectra libraries in mgf or json format (optional)
 
+        If the batch file contains steps requiring input data sets these need to be provided
+        via the optional inputs. Currently this is: 
+
+        - 'Local database' for a 'Custom database search' (LocalCSVDatabaseSearchModule) step 
+
+        Please open an issue if inputs for other steps are needed: https://github.com/galaxyproject/tools-iuc/issues/new
+
         **Outputs**
 
         Galaxy will store all outputs of MZMine in a single collection. When specifying the batch XML
@@ -126,4 +183,4 @@
     <citations>
         <citation type="doi">10.1038/s41587-023-01690-2</citation>
     </citations>
-</tool>
+</tool>
diff --git a/tools/mzmine/set_path.py b/tools/mzmine/set_path.py
@@ -0,0 +1,31 @@
+import argparse
+import re
+import sys
+from xml.etree import ElementTree
+
+parser = argparse.ArgumentParser(
+    prog="set_path.py", description="update paths in mzmine batch XML"
+)
+parser.add_argument("--input", help="input XML")
+parser.add_argument("--output", help="output XML")
+parser.add_argument("--localdb", required=False, help="Local CVS DB for Search")
+args = parser.parse_args()
+
+PATHSEP = re.compile(r"[/\\]")
+tree = ElementTree.parse(args.input)
+
+for batchstep in tree.findall(".//batchstep"):
+    method = batchstep.attrib.get("method")
+    for current_file in batchstep.findall(".//current_file"):
+        if (
+            method
+            == "io.github.mzmine.modules.dataprocessing.id_localcsvsearch.LocalCSVDatabaseSearchModule"
+        ):
+            if args.localdb:
+                current_file.text = args.localdb
+            else:
+                sys.exit("Batch file contains LocalCSVDatabaseSearchModule but no local DB CSV file given")
+        else:
+            current_file.text = f"./output/{PATHSEP.split(current_file.text)[-1]}"
+
+tree.write(args.output)