Sylph tool Wrapper

tools/sylph/.shed.yml

@@ -0,0 +1,13 @@
+name: sylph
+owner: bgruening
+description: sylph - fast and precise species-level metagenomic profiling with ANIs
+long_description: sylph is a program that performs ultrafast (1) ANI querying or (2) metagenomic profiling for metagenomic shotgun samples.
+homepage_url: https://github.com/bluenote-1577/sylph
+remote_repository_url: https://github.com/bgruening/galaxytools/main/tools/sylph
+categories:
+  - Metagenomics
+type: unrestricted
+auto_tool_repositories:
+  name_template: "{{ tool_id }}"
+  description_template: "{{ tool_name }} from the sylph suite"
+

tools/sylph/README.md

@@ -0,0 +1,10 @@
+For Galaxy admins and local runs:
+
+The databases for sylph have associated metadata files. These files MUST be paired with the correct databases to output correctly. Here is the easiest location to download databases and metadata files:
+For databases: https://github.com/bluenote-1577/sylph/wiki/Pre%E2%80%90built-databases
+For metadata: https://github.com/bluenote-1577/sylph-utils
+
+The tool assumes the directory the data_table references to be
+<name_of_organism>
+    - database.syldb
+    - metadata.tsv.gz

tools/sylph/macros.xml

@@ -0,0 +1,147 @@
+<macros>
+    <token name="@TOOL_VERSION@">0.6.1</token>
+    <token name="@VERSION_SUFFIX@">0</token>
+    <token name="@LICENSE@">MIT</token>
+    <token name="@DB_SELECTOR@"><![CDATA[
+        #if $database_select.select == 'cached':
+            ln -s '$database_select.sylph_database.fields.path/database.syldb' 'database.syldb' &&
+        #else:
+            ln -s '$database_select.sylph_database' 'database.syldb' &&
+        #end if
+    ]]></token>
+    <token name="@SINGLE_INPUT@"><![CDATA[
+        #if $sketch.input.ext == 'fastqsanger'
+            #set $ext = 'fastq'
+        #else if $sketch.input.ext == 'fastqsanger.gz':
+            #set $ext = 'fastq.gz'
+        #else:
+            #set $ext = str($sketch.input.ext)
+        #end if
+        #if $sketch.input.element_identifier.endswith('.fastq') or $sketch.input.element_identifier.endswith('.fastq.gz'):
+            #set $input = re.sub(r'\s+', '_', $sketch.input.element_identifier)
+        #else:
+            #set $input = re.sub(r'\s+', '_', $sketch.input.element_identifier + '.' + str($ext))
+        #end if
+        ln -s '$sketch.input' '$input' &&
+    ]]></token>
+    <token name="@SINGLE_GROUP@"><![CDATA[
+        #set input = ''
+        #for $number, $current_file in enumerate($sketch.input):
+            #if $current_file.ext == 'fastqsanger'
+                #set $ext = 'fastq'
+            #else if $current_file.ext == 'fastqsanger.gz':
+                #set $ext = 'fastq.gz'
+            #else:
+                #set $ext = str($current_file.ext)
+            #end if
+            #if $current_file.element_identifier.endswith('.fastq') or $current_file.element_identifier.endswith('.fastq.gz'):
+                #set $current_input = re.sub(r'\s+', '_', $current_file.element_identifier)
+            #else:
+                #set $current_input = re.sub(r'\s+', '_', $current_file.element_identifier + '.' + str($ext))
+            #end if
+            ln -s '${current_file}' '$current_input' &&
+            #set input = str($input) + ' ' + str($current_input)
+        #end for
+    ]]></token>
+    <token name="@PAIRED@"><![CDATA[
+        #if $sketch.input_1.ext == 'fastqsanger'
+                #set $ext_1 = 'fastq'
+            #else if $sketch.input_1.ext == 'fastqsanger.gz':
+                #set $ext_1 = 'fastq.gz'
+            #else:
+                #set $ext_1 = str($sketch.input_1.ext)
+            #end if
+
+            #if $sketch.input_2.ext == 'fastqsanger'
+                #set $ext_2 = 'fastq'
+            #else if $sketch.input_2.ext == 'fastqsanger.gz':
+                #set $ext_2 = 'fastq.gz'
+            #else:
+                #set $ext_2 = str($sketch.input_2.ext)
+            #end if
+
+            #if $sketch.input_1.element_identifier.endswith('.fastq') or $sketch.input_1.element_identifier.endswith('.fastq.gz'):
+                #set $read1 = re.sub(r'\s+', '_', $sketch.input_1.element_identifier)
+            #else:
+                #set $read1 = re.sub(r'\s+', '_', str($sketch.input_1.element_identifier) + '.' + str($ext_1))
+            #end if
+            #if $sketch.input_2.element_identifier.endswith('.fastq') or $sketch.input_2.element_identifier.endswith('.fastq.gz'):
+                #set $read2 = re.sub(r'\s+', '_', $sketch.input_2.element_identifier)
+            #else:
+                #set $read2 = re.sub(r'\s+', '_', str($sketch.input_2.element_identifier) + '.' + str($ext_2))
+            #end if
+            ln -s '$sketch.input_1' '$read1' &&
+            ln -s '$sketch.input_2' '$read2' &&
+    ]]></token>
+    <token name="@PAIRED_GROUP@"><![CDATA[
+        #if $sketch.input.forward.ext == 'fastqsanger'
+            #set $ext_1 = 'fastq'
+        #else if $sketch.input.forward.ext == 'fastqsanger.gz':
+            #set $ext_1 = 'fastq.gz'
+        #else:
+            #set $ext_1 = str($sketch.input.forward.ext)
+        #end if
+
+        #if $sketch.input.reverse.ext == 'fastqsanger'
+            #set $ext_2 = 'fastq'
+        #else if $sketch.input.reverse.ext == 'fastqsanger.gz':
+            #set $ext_2 = 'fastq.gz'
+        #else:
+            #set $ext_2 = str($sketch.input.reverse.ext)
+        #end if
+
+        #set $read1 = re.sub(r'\s+', '_', str($sketch.input.element_identifier) + '.' + str($ext_1))
+        #set $read2 = re.sub(r'\s+', '_', str($sketch.input.element_identifier) + '_r2.' + str($ext_2))
+        ln -s '$sketch.input.forward' '$read1' &&
+        ln -s '$sketch.input.reverse' '$read2' &&
+    ]]></token>
+    <xml name="requirements">
+        <requirements>
+            <requirement type="package" version="@TOOL_VERSION@">sylph</requirement>
+            <requirement type="package" version="3.11">python</requirement>
+            <requirement type="package" version="2.2">pandas</requirement>
+        </requirements>
+    </xml>
+    <xml name="citation">
+        <citations>
+            <citation type="doi">10.1038/s41587-024-02412-y</citation>
+        </citations>
+    </xml>
+    <xml name="creator">
+        <creator>
+            <person givenName="Tyler" familyName="Collins"/>
+            <person givenName="Alexander" familyName="Ostrovsky"/>
+        </creator>
+    </xml>
+    <xml name="xrefs">
+        <xrefs>
+            <xref type="bio.tools">sylph</xref>
+        </xrefs>
+    </xml>
+    <xml name="input_database">
+        <conditional name="database_select">
+            <param name="select" type="select" label="Choose the source for databases and metadata">
+                <option value="cached">Cached data</option>
+                <option value="history">History</option>
+            </param>
+            <when value="cached">
+                <param label="Select a sylph database" name="sylph_database" type="select">
+                    <options from_data_table="sylph_databases">
+                        <!-- <filter type="sort_by" column="3"/> -->
+                        <validator message="No Sylph databases are available" type="no_options" />
+                    </options>
+                </param>
+            </when>
+            <when value="history">
+                <param label="Select a history dataset" name="sylph_database" type="data" format="binary" />
+                <param label="Metadata file for metaphlan and krona outputs" name="metadata" type="data" format="tabular.gz" optional="true" help="The metata file MUST be directly associated with the input database. For more information, view the help text of the tool."/>
+            </when>
+        </conditional>
+    </xml>
+    <xml name="output_format">
+        <param label="Additional output formats" name="outputs" type="select" display="checkboxes" multiple="true" help="In addition to Sylph's tabular output, you may ouput a file converted to these formats">
+            <option value="metaphlan">Sylph's MetaPhlAn-like output</option>
+            <option value="krona">Krona compatible</option>
+        </param>
+    </xml>
+</macros>