annotate meryl.xml @ 9:eadfd71dde37 draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/meryl commit b036dca639d7d67b55929be5575cab71473f53e4"

<tool id='meryl' name='Meryl' version='@TOOL_VERSION@+@GALAXY_TOOL_VERSION@@SUFFIX_VERSION@' profile='20.01'>
    <description>a genomic k-mer counter and sequence utility</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro='edam_ontology' />
    <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 $operation_type.command_type == 'count-kmers'
            #if $operation_type.options_kmer_size.kmer_size == 'estimate'
                #from math import log
                #set size=int(log(int($operation_type.options_kmer_size.genome_size)*(1-float($operation_type.options_kmer_size.collision_rate))/float($operation_type.options_kmer_size.collision_rate))/log(4))
            #elif $operation_type.options_kmer_size.kmer_size == 'provide'
                #set size=$operation_type.options_kmer_size.input_kmer_size
            #end if
            ln -s '$operation_type.input_reads' ./input.${operation_type.input_reads.ext} &&

            meryl
            $operation_type.count_operations
            k=$size
            memory=\$GALAXY_MEMORY_GB
            threads=\${GALAXY_SLOTS:-1}
            ./input.${operation_type.input_reads.ext}
            output read-db.meryl &&
            echo 'K-mer size: ${size}' &&
            tar -zcf read-db.meryldb read-db.meryl
        #elif $operation_type.command_type == 'filter-kmers'
            mkdir  -p ./temp_db/ &&
            tar -zxf $operation_type.input_meryldb_02 -C ./temp_db/ &&
            mv ./temp_db/* tmp.meryl &&
            meryl
            $operation_type.filter_operations
            #if $operation_type.filter_type.type == 'times'
                $operation_type.filter_type.N
            #elif $operation_type.filter_type.type == 'frequency'
                distinct=${operation_type.filter_type.distinct}
            #end if
            tmp.meryl
            output read-db.meryl &&
            tar -zcf read-db.meryldb read-db.meryl
        #elif $operation_type.command_type == 'arithmetic-kmers'
            mkdir  -p ./temp_db/ &&
            tar -zxf $operation_type.input_meryldb_02 -C ./temp_db/ &&
            mv ./temp_db/* tmp.meryl &&
            meryl
            $operation_type.arithmetic_operations
            $operation_type.X
            tmp.meryl
            output read-db.meryl &&
            tar -zcf read-db.meryldb read-db.meryl
        #elif $operation_type.command_type == 'groups-kmers'
            #for $i,$mdb in enumerate($operation_type.input_meryldb_02)
                mkdir -p ./tmp_folder_$i/ &&
                tar -zxf $mdb -C ./tmp_folder_$i &&
                mv ./tmp_folder_$i/* db_'${i}'.meryl &&
            #end for
            meryl $operation_type.groups_operations output read-db.meryl db_*  &&
            tar -zcf read-db.meryldb read-db.meryl
        #elif $operation_type.command_type == 'histogram-kmers'
            mkdir  -p ./temp_db/ &&
            tar -zxf $operation_type.input_meryldb_02 -C ./temp_db/ &&
            mv ./temp_db/* tmp.meryl &&
            meryl histogram tmp.meryl > read-db.hist
        #elif $operation_type.command_type == 'print'
            mkdir  -p ./temp_db/ &&
            tar -zxf $operation_type.input_meryldb_02 -C ./temp_db/ &&
            mv ./temp_db/* tmp.meryl &&
            meryl print tmp.meryl > read-db.tabular
        #elif $operation_type.command_type == 'trio-mode'
            export MERQURY=\$(dirname \$(command -v merqury.sh))/../share/merqury/ &&
            #if $operation_type.options_kmer_size.kmer_size == 'estimate'
                #from math import log
                #set size=int(log(int($operation_type.options_kmer_size.genome_size)*(1-float($operation_type.options_kmer_size.collision_rate))/float($operation_type.options_kmer_size.collision_rate))/log(4))
            #elif $operation_type.options_kmer_size.kmer_size == 'provide'
                #set size=$operation_type.options_kmer_size.input_kmer_size
            #end if
            #for $i, $read in enumerate($paternal_reads):
                mkdir 'paternal{$i}.meryl' &&
                meryl count k=$size memory=\$GALAXY_MEMORY_GB threads=\${GALAXY_SLOTS:-1} '${read}' output 'paternal{$i}.meryl' &&
            #end for
            meryl union-sum paternal*.meryl output pat.meryl &&
            #for $i, $read in enumerate($maternal_reads):
                mkdir 'maternal{$i}.meryl' &&
                meryl count k=$size memory=\$GALAXY_MEMORY_GB threads=\${GALAXY_SLOTS:-1} '${read}' output 'maternal{$i}.meryl' &&
            #end for
            meryl union-sum maternal*.meryl output mat.meryl &&
            #for $i, $read in enumerate($child_reads):
                mkdir 'child{$i}.meryl' &&
                meryl count k=$size memory=\$GALAXY_MEMORY_GB threads=\${GALAXY_SLOTS:-1} '${read}' output 'child{$i}.meryl' &&
            #end for
            meryl union-sum child*.meryl output 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 &&

            ## shared hapmers
            meryl intersect output shared.inherited.meryl child.meryl shared.meryl &&
            meryl histogram shared.inherited.meryl > shared.inherited.hist &&
            java -jar -Xmx1g \$MERQURY/eval/kmerHistToPloidyDepth.jar shared.inherited.hist > shared.inherited.ploidy &&
            VAR=`sed -n 2p shared.inherited.ploidy | awk '{print \$NF}'` &&
            meryl greater-than \$VAR output shared.filt.meryl shared.inherited.meryl &&
            
            ## child hapmers
            meryl union-sum output child.inherited.meryl mat.inherited.meryl pat.inherited.meryl shared.inherited.meryl &&
            meryl difference output read.only.meryl child.meryl child.inherited.meryl &&
            tar -czf 'read-db.meryldb' read.only.meryl &&
            echo 'K-mer size: ${size}'
        #end if
    ]]>    </command>
    <inputs>
        <conditional name="operation_type">
            <param name="command_type" type="select" label="Operation type selector" help="Select a type of operation">
                <option value="count-kmers">Count operations</option>
                <option value="filter-kmers">Filter operations</option>
                <option value="arithmetic-kmers">Arithmetic operations on k-mer counts</option>
                <option value="groups-kmers">Operations on sets of k-mers</option>
                <option value="histogram-kmers">Generate histogram dataset</option>
                <option value="print">Print k-mer counts to a tabular file</option>
                <option value="trio-mode">Build hap-mer dbs for trios</option>
            </param>
            <when value="count-kmers">
                <param name="count_operations" 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>
            </when>
            <when value="filter-kmers">
                <param name="filter_operations" type="select" label="Filter operations" help="Select an operation to be executed">
                    <option value="less-than">Less-than: return k-mers that occur fewer than a threshold value</option>
                    <option value="greater-than">Greater-than: return k-mers that occur more than a threshold value</option>
                    <option value="equal-to">Equal-to: return k-mers that occur exactly a threshold value</option>
                    <option value="not-equal-to">Not-equal-to: return k-mers that do not occur exactly a threshold value</option>
                </param>
                <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb" help="Select a meryldb dataset" />

                <conditional name="filter_type">
                    <param name="type" type="select" label="Type of filtering">
                        <option value="times">Return k-mers that occur N times in the input</option>
                        <option value="frequency">Return k-mers occurring at specific frequencies</option>
                    </param>
                    <when value="times">
                        <param name="N" type="integer" min="0" max="50000000000" value="" optional="true" label="Number of times in the input" help="Return k-mers that occur N times in the input." />
                    </when>
                    <when value="frequency">
                        <param name="distinct" type="float" min="0" max="1" value="0.9998" optional="true" label="Frequency" help="Return k-mers that at specific frequency (e.g. frequency = 0.9998 returns top 0.02% most frequent)" />
                    </when>
                </conditional>
            </when>
            <when value="arithmetic-kmers">
                <param name="arithmetic_operations" type="select" label="Arithmetic operations" help="Select an operation to be executed">
                    <option value="increase">Increase: add x to the count of each k-mer</option>
                    <option value="decrease">Decrease: subtract x from the count of each k-mer</option>
                    <option value="multiply">Multiply: multiply the count of each k-mer by x</option>
                    <option value="divide">Divide: divide the count of each k-mer by x</option>
                    <option value="divide-round">Divide-round: divide the count of each k-mer by x and round th results</option>
                    <option value="modulo">Modulo: set the count of each k-mer to the remainder of the count divided by x</option>
                </param>
                <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb" help="Select a meryldb dataset" />
                <param name="X" type="integer" min="1" max="1000000" value="" optional="true" label="Operand" />
            </when>
            <when value="groups-kmers">
                <param name="groups_operations" type="select" label="Operations on sets of k-mers" help="Select an operation to be executed">
                    <option value="union">Union: return k-mers that occur in any input</option>
                    <option value="union-min">Union-min: return k-mers that occur in any input, set the count to the minimum count</option>
                    <option value="union-max">Union-max: return k-mers that occur in any input, set the count to the maximum count</option>
                    <option value="union-sum">Union-sum: return k-mers that occur in any input, set the count to the sum of the counts</option>
                    <option value="intersect">Intersect: return k-mers that occur in all inputs, set the count to the count in the first input</option>
                    <option value="intersect-min">Intersect-min: return k-mers that occur in all inputs, set the count to the minimum count</option>
                    <option value="intersect-max">Intersect-max: return k-mers that occur in all inputs, set the count to the maximum count</option>
                    <option value="intersect-sum">Intersect-sum: return k-mers that occur in all inputs, set the count to the sum of the counts</option>
                    <option value="subtract">Subtract: return k-mers that occur in the first input, subtracting counts from the other inputs</option>
                    <option value="difference">Difference: return k-mers that occur in the first input, but none of the other inputs</option>
                    <option value="symmetric-difference">Symmetric-difference: return k-mers that occur in exactly one input</option>
                </param>
                <param name="input_meryldb_02" type="data" multiple="true" format="meryldb" label="Input meryldb" />
            </when>
            <when value="histogram-kmers">
                <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb" />
            </when>
            <when value="print">
                <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb" />
            </when>
            <when value="trio-mode">
                <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>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb" label="${tool.name} on ${on_string}: read-db.meryldb">
            <filter>operation_type["command_type"] != "histogram-kmers" and operation_type["command_type"] != "print"</filter>
        </data>
        <data name="read_db_hist" format="tabular" from_work_dir="read-db.hist" label="${tool.name} on ${on_string}: read-db histogram">
            <filter>operation_type["command_type"] == "histogram-kmers" or operation_type["command_type"] == "trio-mode"</filter>
        </data>
        <data name="read_db_print" format="tabular" from_work_dir="read-db.tabular" label="${tool.name} on ${on_string}: kmer counts">
            <filter>operation_type["command_type"] == "print"</filter>
        </data>
        <data name="pat_db" format="meryldb" from_work_dir="pat.meryldb" label="${tool.name} on ${on_string}: pat.meryldb">
            <filter>operation_type["command_type"] == "trio-mode"</filter>
        </data>
        <data name="pat_db_hist" format="tabular" from_work_dir="pat.inherited.hist" label="${tool.name} on ${on_string}: paternal inherited histogram">
            <filter>operation_type["command_type"] == "trio-mode"</filter>
        </data>
        <data name="mat_db" format="meryldb" from_work_dir="mat.meryldb" label="${tool.name} on ${on_string}: mat.meryldb">
            <filter>operation_type["command_type"] == "trio-mode"</filter>
        </data>
        <data name="mat_db_hist" format="tabular" from_work_dir="mat.inherited.hist" label="${tool.name} on ${on_string}: maternal inherited histogram">
            <filter>operation_type["command_type"] == "trio-mode"</filter>
        </data>
    </outputs>
    <tests>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="count-kmers" />
                <param name="count_operation" value="count" />
                <param name="input_reads" value="child.fasta" />
                <conditional name="options_kmer_size">
                    <param name="kmer_size" value="provide" />
                    <param name="input_kmer_size" value="7" />
                </conditional>
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="22152" delta="1000" />
                </assert_contents>
            </output>
            <assert_stdout>
                <has_line line="K-mer size: 7" />
            </assert_stdout>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="filter-kmers" />
                <param name="filter_operations" value="less-than" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
                <conditional name="filter_type">
                    <param name="N" value="100" />
                </conditional>
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="32077" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="filter-kmers" />
                <param name="filter_operations" value="greater-than" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
                <conditional name="filter_type">
                    <param name="N" value="80" />
                </conditional>
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="49643" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="filter-kmers" />
                <param name="filter_operations" value="greater-than" />
                <param name="input_meryldb_02" value="maternal.meryldb" ftype="meryldb" />
                <conditional name="filter_type">
                    <param name="distinct" value="0.9998" />
                </conditional>
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="1634" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="filter-kmers" />
                <param name="filter_operations" value="equal-to" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
                <conditional name="filter_type">
                    <param name="N" value="100" />
                </conditional>
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="2621" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="filter-kmers" />
                <param name="filter_operations" value="not-equal-to" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
                <conditional name="filter_type">
                    <param name="N" value="100" />
                </conditional>
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="59100" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="arithmetic-kmers" />
                <param name="arithmetic_operations" value="increase" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
                <param name="X" value="100000" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="59500" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="arithmetic-kmers" />
                <param name="arithmetic_operations" value="decrease" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
                <param name="X" value="100" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="42313" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="arithmetic-kmers" />
                <param name="arithmetic_operations" value="multiply" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
                <param name="X" value="3" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="60530" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="arithmetic-kmers" />
                <param name="arithmetic_operations" value="divide" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
                <param name="X" value="2" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="56200" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="arithmetic-kmers" />
                <param name="arithmetic_operations" value="divide-round" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
                <param name="X" value="2" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="56100" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <!-- test 12 -->
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="arithmetic-kmers" />
                <param name="arithmetic_operations" value="modulo" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
                <param name="X" value="3" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="37501" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="union" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
                <param name="input_meryldb_03" value="" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="36100" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="union-min" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="58925" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="union-max" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="58930" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="union-sum" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="58600" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="intersect" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="14951" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="intersect-min" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="14957" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="intersect-max" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="14956" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="intersect-sum" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="14953" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <!-- test 20 -->
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="subtract" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="23999" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="difference" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="24016" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="groups-kmers" />
                <param name="groups_operations" value="symmetric-difference" />
                <param name="input_meryldb_02" value="output_02.read-db.meryldb,output_03.read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="57455" delta="1000" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="6">
            <conditional name="operation_type">
                <param name="command_type" value="trio-mode" />
                <param name="child_reads" value="child.fasta" />
                <param name="paternal_reads" value="paternal.fasta" />
                <param name="maternal_reads" value="maternal.fasta" />
                <conditional name="options_kmer_size">
                    <param name="kmer_size" value="provide" />
                    <param name="input_kmer_size" value="7" />
                </conditional>
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="1573" delta="300" />
                </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="1779" delta="300" />
                </assert_contents>
            </output>
            <output name="pat_db_hist" file="output_23.pat.hist" />
            <output name="mat_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="1569" delta="300" />
                </assert_contents>
            </output>
            <output name="mat_db_hist" file="output_23.mat.hist" />
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="histogram-kmers" />
                <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db_hist" file="output_24.read-db.hist" />
        </test>
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="print" />
                <param name="input_meryldb_02" value="maternal.meryldb" ftype="meryldb" />
            </conditional>
            <output name="read_db_print" file="output_25.read-db.tabular" sort="true" />
        </test>
        <!--Test compressed files-->
        <test expect_num_outputs="1">
            <conditional name="operation_type">
                <param name="command_type" value="count-kmers" />
                <param name="count_operation" value="count" />
                <param name="input_reads" value="child.fasta.gz" ftype="fasta.gz" />
                <conditional name="options_kmer_size">
                    <param name="kmer_size" value="provide" />
                    <param name="input_kmer_size" value="7" />
                </conditional>
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="22200" delta="1000" />
                </assert_contents>
            </output>
            <assert_stdout>
                <has_line line="K-mer size: 7" />
            </assert_stdout>
        </test>
    </tests>
    <help><![CDATA[

.. class:: infomark

**Purpose**

Meryl is the k-mer counter. It is built into the Celera assembler and is also available as a stand-alone application.
Meryl uses a sorting-based approach that sorts the k-mers in lexicographical order.

In addition of generating count-databases, meryl can perform simple operations on it. 

-----

.. class:: infomark

**Basic functions**

The functions that meryl includes are described below: 

    ::

        COUNT OPERATIONS
        - 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
        FILTERING OPERATIONS
        - Less-than: return k-mers that occur fewer than N times in the input
        - Greater-than: return k-mers that occur more than N times in the input
        - Equal-to: return k-mers that occur exactly N times in the input
        - Not-equal-to: return k-mers that do not occur exactly N times in the input
        ARITHMETIC OPERATIONS
        - Increase: add x to the count of each k-mer
        - Decrease: subsctract x from the count of each k-mer
        - Multiply: multiply the count of each k-mer by x
        - Divide: divide the count of each k-mer by x
        - Divide-round: divide the count of each k-mer by x and round th results
        - Modulo: set the count of each k-mer to the remainder of the count divided by x
        OPERATIONS ON SETS
        - Union-min: return k-mers that occur in any input, set the count to the minimum count
        - Union-max: return k-mers that occur in any input, set the count to the maximum count
        - Union-sum: return k-mers that occur in any input, set the count to the sum of the counts
        - Intersect: return k-mers that occur in all inputs, set the count to the count in the first input
        - Intersect-min: return k-mers that occur in all inputs, set the count to the minimum count
        - Intersect-max: return k-mers that occur in all inputs, set the count to the maximum count
        - Intersect-sum: return k-mers that occur in all inputs, set the count to the sum of the counts
        - Subtract: return k-mers that occur in the first input, subtracting counts from the other inputs
        - Difference: return k-mers that occur in the first input, but none of the other inputs
        - Symmetric-difference: return k-mers that occur in exactly one input

-----

.. class:: infomark

**Additional function: build hap-mers dbs for trios**

In addition of the basic operations, this wrapper allows to build the hap-mers 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>