Mercurial > repos > iuc > meryl_trio_mode
diff trio-mode.xml @ 0:c85aec7804a4 draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/meryl commit a95f5b04b1219489a327a622184633a561fe5ac0
| author | iuc |
|---|---|
| date | Thu, 25 Apr 2024 21:01:24 +0000 |
| parents | |
| children |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/trio-mode.xml Thu Apr 25 21:01:24 2024 +0000 @@ -0,0 +1,146 @@ +<tool id="meryl_trio_mode" name="Meryl" version="@TOOL_VERSION@+@GALAXY_TOOL_VERSION@@SUFFIX_VERSION@" profile="@PROFILE@"> + <description>build hap-mers databases for trios</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)) && + export MERQURY=\$(dirname \$(command -v merqury.sh))/../share/merqury/ && + #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 + #for $i, $read in enumerate($paternal_reads): + mkdir 'paternal${i}.meryl' && + ln -s '$read' ./input_paternal_${i}.${read.ext} && + meryl count k=$size memory=\$GALAXY_MEMORY_GB threads=\${GALAXY_SLOTS:-1} ./input_paternal_${i}.${read.ext} output 'paternal${i}.meryl' && + #end for + meryl union-sum paternal*.meryl output pat.meryl && + #for $i, $read in enumerate($maternal_reads): + mkdir 'maternal${i}.meryl' && + ln -s '$read' ./input_maternal_${i}.${read.ext} && + meryl count k=$size memory=\$GALAXY_MEMORY_GB threads=\${GALAXY_SLOTS:-1} ./input_maternal_${i}.${read.ext} output 'maternal${i}.meryl' && + #end for + meryl union-sum maternal*.meryl output mat.meryl && + #for $i, $read in enumerate($child_reads): + mkdir 'child${i}.meryl' && + ln -s '$read' ./input_child_${i}.${read.ext} && + meryl count k=$size memory=\$GALAXY_MEMORY_GB threads=\${GALAXY_SLOTS:-1} ./input_child_${i}.${read.ext} output 'child${i}.meryl' && + #end for + meryl union-sum child*.meryl output child.meryl && + meryl histogram child.meryl > read-db.hist && + tar -czf 'read-db.meryldb' child.meryl && + + ## mat specific kmers + meryl difference mat.meryl pat.meryl output mat.only.meryl && + meryl histogram mat.only.meryl > mat.only.hist && + java -jar -Xmx1g \$MERQURY/eval/kmerHistToPloidyDepth.jar mat.only.hist > mat.only.ploidy && + VAR=`sed -n 2p mat.only.ploidy | awk '{print \$NF}'` && + meryl greater-than \$VAR output mat.only.filt.meryl mat.only.meryl && + + ## pat specific kmers + meryl difference pat.meryl mat.meryl output pat.only.meryl && + meryl histogram pat.only.meryl > pat.only.hist && + java -jar -Xmx1g \$MERQURY/eval/kmerHistToPloidyDepth.jar pat.only.hist > pat.only.ploidy && + VAR=`sed -n 2p pat.only.ploidy | awk '{print \$NF}'` && + meryl greater-than \$VAR output pat.only.filt.meryl pat.only.meryl && + + ## shared kmers + meryl intersect output shared.meryl mat.meryl pat.meryl && + + ## mat hapmers + meryl intersect output mat.inherited.meryl child.meryl mat.only.filt.meryl && + meryl histogram mat.inherited.meryl > mat.inherited.hist && + java -jar -Xmx1g \$MERQURY/eval/kmerHistToPloidyDepth.jar mat.inherited.hist > mat.inherited.ploidy && + VAR=`sed -n 2p mat.inherited.ploidy | awk '{print \$NF}'` && + meryl greater-than \$VAR output mat.hapmer.meryl mat.inherited.meryl && + tar -czf 'mat.meryldb' mat.hapmer.meryl && + + ## pat hapmers + meryl intersect output pat.inherited.meryl child.meryl pat.only.filt.meryl && + meryl histogram pat.inherited.meryl > pat.inherited.hist && + java -jar -Xmx1g \$MERQURY/eval/kmerHistToPloidyDepth.jar pat.inherited.hist > pat.inherited.ploidy && + VAR=`sed -n 2p pat.inherited.ploidy | awk '{print \$NF}'` && + meryl greater-than \$VAR output pat.hapmer.meryl pat.inherited.meryl && + tar -czf 'pat.meryldb' pat.hapmer.meryl && + + echo 'K-mer size: ${size}' + ]]></command> + <inputs> + <param name="child_reads" type="data" format="fastq,fasta,fastq.gz,fasta.gz" multiple="true" label="F1 reads" help="Select F1 reads used for generating the assembly"/> + <param name="paternal_reads" type="data" format="fastq,fasta,fastq.gz,fasta.gz" multiple="true" optional="true" label="Paternal reads" help="Select the paternal reads used for generating the assembly"/> + <param name="maternal_reads" type="data" format="fastq,fasta,fastq.gz,fasta.gz" multiple="true" optional="true" label="Maternal reads" help="Select the maternal reads used for generating the assembly"/> + <conditional name="options_kmer_size"> + <param name="kmer_size" type="select" label="K-mer size selector"> + <option value="provide">Set a k-mer size</option> + <option value="estimate">Estimate the best k-mer size</option> + </param> + <when value="provide"> + <param name="input_kmer_size" type="integer" min="1" max="50" value="" label="K-mer size" help="For a human genome, the best k-mer size is k=21 for both haploid (3.1G) or diploid (6.2G)."/> + </when> + <when value="estimate"> + <param name="genome_size" type="integer" min="1000" max="70000000000" value="1000" label="Genome size" help="Haploid genome size or diploid genome size, depending on what we evaluate. In bp. Only required if the k-mer size is not provided."/> + <param name="collision_rate" type="float" min="0.0001" max="0.01" value="0.001" label="Tolerable collision rate" help="Tolerable collision rate. By default is 0.001."/> + </when> + </conditional> + </inputs> + <outputs> + <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb" label="${tool.name} on ${on_string}: read-db.meryldb"/> + <data name="read_db_hist" format="tabular" from_work_dir="read-db.hist" label="${tool.name} on ${on_string}: read-db histogram"/> + <data name="pat_db" format="meryldb" from_work_dir="pat.meryldb" label="${tool.name} on ${on_string}: pat.meryldb"/> + <data name="pat_db_hist" format="tabular" from_work_dir="pat.inherited.hist" label="${tool.name} on ${on_string}: paternal inherited histogram"/> + <data name="mat_db" format="meryldb" from_work_dir="mat.meryldb" label="${tool.name} on ${on_string}: mat.meryldb"/> + <data name="mat_db_hist" format="tabular" from_work_dir="mat.inherited.hist" label="${tool.name} on ${on_string}: maternal inherited histogram"/> + </outputs> + <tests> + <test expect_num_outputs="6"> + <conditional name="options_kmer_size"> + <param name="kmer_size" value="provide"/> + <param name="input_kmer_size" value="21"/> + </conditional> + <param name="maternal_reads" value="maternal.fasta"/> + <param name="paternal_reads" value="paternal.fasta"/> + <param name="child_reads" value="child.fasta"/> + <output name="read_db" ftype="meryldb"> + <assert_contents> + <has_size value="3362942" delta="2000"/> + <expand macro="meryldb_archive_assumptions"/> + </assert_contents> + </output> + <output name="read_db_hist" file="output_23.read-db.hist"/> + <output name="pat_db" ftype="meryldb"> + <assert_contents> + <has_size value="120610" delta="400"/> + <expand macro="meryldb_archive_assumptions"/> + </assert_contents> + </output> + <output name="pat_db_hist" file="output_23.pat.hist"/> + <output name="mat_db" ftype="meryldb"> + <assert_contents> + <has_size value="67883" delta="300"/> + <expand macro="meryldb_archive_assumptions"/> + </assert_contents> + </output> + <output name="mat_db_hist" file="output_23.mat.hist"/> + </test> + </tests> + <help> + +.. class:: infomark + +**Purpose** + +Meryl is the k-mer counter. This tool builds hap-mer databases for trios, in accordance +with `merqury's recommended guidelines. <https://github.com/marbl/merqury/wiki/1.-Prepare-meryl-dbs#3-build-hap-mer-dbs-for-trios>`_ + + </help> + <expand macro="citations"/> +</tool>