Merge pull request Helmholtz-UFZ#30 from Helmholtz-UFZ/virsorter

Add virsorter
rmassei · Jun 23, 2024 · fdde667 · fdde667
2 parents 0d3034a + 1818676
commit fdde667
Show file tree

Hide file tree

Showing 13 changed files with 383 additions and 0 deletions.
diff --git a/data_managers/data_manager_virsorter/.shed.yml b/data_managers/data_manager_virsorter/.shed.yml
@@ -0,0 +1,10 @@
+categories:
+- Data Managers
+- Metagenomics
+description: Data managers for virsorter reference data
+homepage_url: https://github.com/jiarong/VirSorter2
+long_description: Data managers for virsorter reference data
+name: virsorter_build_database
+owner: ufz
+remote_repository_url: https://github.com/Helmholtz-UFZ/ufz-galaxy-tools/blob/main/data_managers/data_manager_virsorter
+type: unrestricted
diff --git a/data_managers/data_manager_virsorter/data_manager/virsorter_datamanager.xml b/data_managers/data_manager_virsorter/data_manager/virsorter_datamanager.xml
@@ -0,0 +1,60 @@
+<?xml version="1.0"?>
+<tool id="virsorter_build_database" name="virsorter" tool_type="manage_data" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
+    <description>database builder</description>
+    <macros>
+        <token name="@TOOL_VERSION@">2.2.4</token>  <!-- when updating make sure to double check the db version -->
+        <token name="@VERSION_SUFFIX@">0</token>
+        <token name="@PROFILE@">22.01</token>
+        <!-- zenodo ID and therebut the version of the reference data is hardcoded in the current version virsorter:
+             https://github.com/jiarong/VirSorter2/blob/c0dcac53f3373244bf1bd552a7c2104d53802df7/virsorter/rules/setup.smk#L13 
+             check and update the DB version when updating the VS version 
+             might also change/improve in the future https://github.com/jiarong/VirSorter2/issues/205#issuecomment-2184184977
+             -->
+        <token name="@VIRSORTER_DB_VERSION@">0.4</token>
+    </macros>
+    <requirements>
+        <requirement type="package" version="@TOOL_VERSION@">virsorter</requirement>
+        <requirement type="package" version="1.21.4">wget</requirement>
+        <requirement type="package" version="2.7.0">pulp</requirement> <!-- needs to be pinned, because the old-ish snakemake used in vs2 struggles with https://github.com/snakemake/snakemake/issues/2607 -->
+    </requirements>
+    <command detect_errors="exit_code"><![CDATA[
+        virsorter setup -d db -j \${GALAXY_SLOTS:-1} --skip-deps-install &&
+        cp '$dmjson' '$out_file'
+    ]]></command>
+    <configfiles>
+        <configfile name="dmjson"><![CDATA[
+{
+  "data_tables":{
+    "virsorter":[
+      {
+        "path":"db",
+        "name":"Version @VIRSORTER_DB_VERSION@",
+        "virsorter_version":"@TOOL_VERSION@",
+        "value":"@VIRSORTER_DB_VERSION@"
+      }
+    ]
+  }
+}]]>
+        </configfile>
+    </configfiles>
+    <inputs/>
+    <outputs>
+        <data name="out_file" format="data_manager_json" />
+    </outputs>
+    <tests>
+        <test expect_num_outputs="1">
+            <output name="out_file">
+                <assert_contents>
+                    <has_text text='"value":"@VIRSORTER_DB_VERSION@"'/>
+                    <has_text text='"name":"Version @VIRSORTER_DB_VERSION@"'/>
+                </assert_contents>
+            </output>
+        </test>
+    </tests>
+    <help><![CDATA[
+Download and extract virsorter database version @VIRSORTER_DB_VERSION@ from zenodo (https://zenodo.org/records/10594875)
+    ]]></help>
+    <citations>
+        <citation type="doi">10.1186/s40168-020-00990-y</citation>
+    </citations>
+</tool>
diff --git a/data_managers/data_manager_virsorter/data_manager_conf.xml b/data_managers/data_manager_virsorter/data_manager_conf.xml
@@ -0,0 +1,19 @@
+<data_managers>
+    <data_manager tool_file="data_manager/virsorter_datamanager.xml" id="virsorter_build_database">
+        <data_table name="virsorter">
+            <output>
+                <column name="value"/>
+                <column name="name"/>
+                <column name="virsorter_version"/>
+                <column name="path" output_ref="out_file">
+                    <move type="directory">
+                        <source>${path}</source>
+                        <target base="${GALAXY_DATA_MANAGER_DATA_PATH}">virsorter/${path}</target>
+                    </move>
+                    <value_translation>${GALAXY_DATA_MANAGER_DATA_PATH}/virsorter/${path}</value_translation>
+                    <value_translation type="function">abspath</value_translation>
+                </column>
+            </output>
+        </data_table>
+    </data_manager>
+</data_managers>
diff --git a/data_managers/data_manager_virsorter/tool-data/virsorter.loc.sample b/data_managers/data_manager_virsorter/tool-data/virsorter.loc.sample
@@ -0,0 +1,3 @@
+# Format:
+# value	name	virsorter_version	path
+# value must be the version of the database, e.g. 0.4, needs to be a number
diff --git a/data_managers/data_manager_virsorter/tool_data_table_conf.xml.sample b/data_managers/data_manager_virsorter/tool_data_table_conf.xml.sample
@@ -0,0 +1,8 @@
+<tables>
+    <!-- virsorter reference data  -->
+    <table name="virsorter" comment_char="#">
+        <!-- value should be the version (https://zenodo.org/records/4297575)  -->
+        <columns>value, name, virsorter_version, path</columns>
+        <file path="tool-data/virsorter.loc" />
+    </table>
+</tables>
diff --git a/tools/virsorter/.shed.yml b/tools/virsorter/.shed.yml
@@ -0,0 +1,18 @@
+categories:
+- Metagenomics
+description: VirSorter2 applies a multi-classifier, expert-guided approach to detect diverse DNA and RNA virus genomes.
+long_description: |
+  VirSorter2 applies a multi-classifier, expert-guided approach to detect diverse DNA and RNA virus genomes.
+name: virsorter
+owner: ufz
+homepage_url: https://github.com/jiarong/VirSorter2/
+remote_repository_url: https://github.com/Helmholtz-UFZ/ufz-galaxy-tools/blob/main/tools/virsorter
+type: unrestricted
+auto_tool_repositories:
+  name_template: "{{ tool_id }}"
+  description_template: "Wrapper for virsorter function: {{ tool_name }}."
+suite:
+  name: "suite_virsorter"
+  description: "A suite of tools that brings the virsorter project into Galaxy."
+  long_description: |
+    VirSorter2 applies a multi-classifier, expert-guided approach to detect diverse DNA and RNA virus genomes.
diff --git a/tools/virsorter/test-data.sh b/tools/virsorter/test-data.sh
@@ -0,0 +1,4 @@
+#!/bin/bash
+
+cd test-data/
+wget -O - https://osf.io/v46sc/download | tar -xvz 
diff --git a/tools/virsorter/test-data/8seq.fa b/tools/virsorter/test-data/8seq.fa
diff --git a/tools/virsorter/test-data/virsorter.loc b/tools/virsorter/test-data/virsorter.loc
@@ -0,0 +1,3 @@
+# value	name	path
+# value must be the version of the database, e.g. 0.4, needs to be a number
+0.4	virsorter 0.4	2.2.4	${__HERE__}/db/
diff --git a/tools/virsorter/tool-data/virsorter.loc.sample b/tools/virsorter/tool-data/virsorter.loc.sample
@@ -0,0 +1,3 @@
+# Your virsorter.loc file should include an entry per line
+# value should be db version, i.e. a number
+#value	name	virsorter_version	path
diff --git a/tools/virsorter/tool_data_table_conf.xml.sample b/tools/virsorter/tool_data_table_conf.xml.sample
@@ -0,0 +1,10 @@
+<?xml version="1.0"?>
+<tables>
+    <!-- Locations of indexes for virsorter, since reference data seems not to be versioned we use the download date as version -->
+    <table name="virsorter" comment_char="#">
+        <columns>value, name, virsorter_version, path</columns>
+        <file path="tool-data/virsorter.loc" />
+    </table>
+</tables>
+
+
diff --git a/tools/virsorter/tool_data_table_conf.xml.test b/tools/virsorter/tool_data_table_conf.xml.test
@@ -0,0 +1,6 @@
+<tables>
+    <table name="virsorter" comment_char="#">
+        <columns>value, name, virsorter_version, path</columns>
+        <file path="${__HERE__}/test-data/virsorter.loc" />
+    </table>
+</tables>