--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/count-kmers.xml	Tue Oct 08 16:42:39 2024 +0000
@@ -0,0 +1,122 @@
+<tool id="meryl_count_kmers" name="Meryl" version="@TOOL_VERSION@+@GALAXY_TOOL_VERSION@@SUFFIX_VERSION@" profile="@PROFILE@">
+    <description>count k-mers</description>
+    <macros>
+        <import>macros.xml</import>
+    </macros>
+    <expand macro="edam_ontology"/>
+    <xrefs>
+        <xref type="bio.tools">meryl</xref>
+    </xrefs>
+    <expand macro="requirements"/>
+    <version_command>meryl --version</version_command>
+    <command detect_errors="exit_code"><![CDATA[
+    export GALAXY_MEMORY_GB=\$((\${GALAXY_MEMORY_MB:-8192}/1024)) &&
+    #if $options_kmer_size.kmer_size == 'estimate'
+        #from math import log
+        #set size=int(log(int($options_kmer_size.genome_size)*(1-float($options_kmer_size.collision_rate))/float($options_kmer_size.collision_rate))/log(4))
+    #elif $options_kmer_size.kmer_size == 'provide'
+        #set size=$options_kmer_size.input_kmer_size
+    #end if
+    ln -s '$input_reads' ./input.${input_reads.ext} &&
+
+    meryl
+    $count_operation
+    k=$size
+    memory=\$GALAXY_MEMORY_GB
+    threads=\${GALAXY_SLOTS:-1}
+    ./input.${input_reads.ext}
+    output read-db.meryl &&
+    echo 'K-mer size: ${size}' &&
+    tar -cf read-db.meryldb --use-compress-program="pigz -p \${GALAXY_SLOTS:-1}" read-db.meryl
+    ]]></command>
+    <inputs>
+        <param name="count_operation" type="select" label="Count operations" help="Select an operation to be executed">
+            <option value="count">Count: count the occurrences of canonical k-mers</option>
+            <option value="count-forward">Count-forward: count the occurreces of forward k-mers</option>
+            <option value="count-reverse">Count-reverse: count the occurreces of reverse k-mers</option>
+        </param>
+        <param name="input_reads" type="data" format="fastq,fastq.gz,fasta,fasta.gz" label="Input sequences" help="Select your reads in FASTA/FASTQ format."/>
+        <conditional name="options_kmer_size">
+            <param name="kmer_size" type="select" label="K-mer size selector">
+                <option value="provide">Set a k-mer size</option>
+                <option value="estimate">Estimate the best k-mer size</option>
+            </param>
+            <when value="provide">
+                <param name="input_kmer_size" type="integer" min="1" max="50" value="" label="K-mer size" help="For a human genome, the best k-mer size is k=21 for both haploid (3.1G) or diploid (6.2G).."/>
+            </when>
+            <when value="estimate">
+                <param name="genome_size" type="integer" min="1000" max="70000000000" value="1000" label="Genome size" help="Haploid genome size or diploid genome size, depending on what we evaluate. In bp. Only required if the k-mer size is not provided."/>
+                <param name="collision_rate" type="float" min="0.0001" max="0.01" value="0.001" label="Tolerable collision rate" help="Tolerable collision rate. By default is 0.001."/>
+            </when>
+        </conditional>
+    </inputs>
+    <outputs>
+        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb"/>
+    </outputs>
+    <tests>
+        <test expect_num_outputs="1">
+            <conditional name="options_kmer_size">
+                <param name="kmer_size" value="provide"/>
+                <param name="input_kmer_size" value="7"/>
+            </conditional>
+            <param name="input_reads" value="child.fasta"/>
+            <param name="count_operation" value="count"/>
+            <output name="read_db" ftype="meryldb">
+                <assert_contents>
+                    <has_size value="23404" delta="3000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
+                </assert_contents>
+            </output>
+            <assert_stdout>
+                <has_line line="K-mer size: 7"/>
+            </assert_stdout>
+        </test>
+        <test expect_num_outputs="1">
+            <conditional name="options_kmer_size">
+                <param name="kmer_size" value="provide"/>
+                <param name="input_kmer_size" value="7"/>
+            </conditional>
+            <param name="input_reads" value="child.fasta.gz" ftype="fasta.gz"/>
+            <param name="count_operation" value="count"/>
+            <output name="read_db" ftype="meryldb">
+                <assert_contents>
+                    <has_size value="22700" delta="3000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
+                </assert_contents>
+            </output>
+            <assert_stdout>
+                <has_line line="K-mer size: 7"/>
+            </assert_stdout>
+        </test>
+        <test expect_num_outputs="1">
+            <conditional name="options_kmer_size">
+                <param name="kmer_size" value="provide"/>
+                <param name="input_kmer_size" value="7"/>
+            </conditional>
+            <param name="input_reads" value="child.fastq.gz" ftype="fastqsanger.gz"/>
+            <param name="count_operation" value="count"/>
+            <output name="read_db" ftype="meryldb">
+                <assert_contents>
+                    <has_size value="23155" delta="3000"/>
+                    <expand macro="meryldb_archive_assumptions"/>
+                </assert_contents>
+            </output>
+            <assert_stdout>
+                <has_line line="K-mer size: 7"/>
+            </assert_stdout>
+        </test>
+    </tests>
+    <help>
+
+.. class:: infomark
+
+**Purpose**
+
+Meryl is the k-mer counter. This tool can be used to count kmers. 
+
+- Count: count the occurrences of canonical k-mers
+- Count-forward: count the occurreces of forward k-mers
+- Count-reverse: count the occurreces of reverse k-mers
+        </help>
+    <expand macro="citations"/>
+</tool>