comparison meryl.xml @ 4:d5d68989caec draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/meryl commit 51618128e9d1a23f1f3a8c9df7f2b23a2ba3f3c2"

3:9f7a227436d6

            mkdir  -p ./temp_db/ &&
            tar -zxf $operation_type.input_meryldb_02 -C ./temp_db/ &&
            mv ./temp_db/* tmp.meryl &&
            meryl
            $operation_type.filter_operations
            $operation_type.N
            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/ &&

        #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 == '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))

                <option value="count-kmers">Count operations</option>
                <option value="filter-kmers">Filter operations</option>
                <option value="arithmetic-kmers">Arithmetic operations on kmer counts</option>
                <option value="groups-kmers">Operations on sets of k-mers</option>
                <option value="histogram-kmers">Generate histogram dataset</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>

                    </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 N times in the input</option>
                    <option value="greater-than">Greater-than: return k-mers that occur more than N times in the input</option>
                    <option value="equal-to">Equal-to: return k-mers that occur exactly N times in the input</option>
                    <option value="not-equal-to">Not-equal-to: return k-mers that do not occur exactly N times in the input</option>
                </param>
                <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb" help="Select a meryldb dataset" />
                <param name="N" type="integer" min="0" max="50000000000" value="" optional="true" label="Filter value" />
            </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: subsctract x from the count of each k-mer</option>

                <param name="input_meryldb_03" type="data" 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="trio-mode">
                <param name="child_reads" type="data" format="fastq,fasta" multiple="true" label="F1 reads" help="Select F1 reads used for generating the assembly" />
                <param name="paternal_reads" type="data" format="fastq,fasta" 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" multiple="true" optional="true" label="Maternal reads" help="Select the maternal reads used for generating the assembly" />
                <conditional name="options_kmer_size">

            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb" label="${tool.name} on ${on_string}: read-db.meryldb">
            <filter>operation_type["command_type"] != "histogram-kmers"</filter>
        </data>
        <data name="read_db_hist" format="tabular" from_work_dir="read-db.hist" label="${tool.name} on ${on_string}: read-db histogram">
            <filter>operation_type["command_type"] == "histogram-kmers" or operation_type["command_type"] == "trio-mode"</filter>
        </data>
        <data name="pat_db" format="meryldb" from_work_dir="pat.meryldb" label="${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">

        <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" />
                <param name="N" value="100" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="32077" delta="300" />
                </assert_contents>

        <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" />
                <param name="N" value="80" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="49643" delta="300" />
                </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" />
                <param name="N" value="100" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="2621" delta="300"/>
                </assert_contents>

        <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" />
                <param name="N" value="100" />
            </conditional>
            <output name="read_db" ftype="meryldb">
                <assert_contents>
                    <has_size value="59378" delta="300" />
                </assert_contents>

                <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>
    </tests>
    <help><![CDATA[

.. class:: infomark

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: 

    ::

        - 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>`_

4:d5d68989caec

            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/ &&

        #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))

                <option value="count-kmers">Count operations</option>
                <option value="filter-kmers">Filter operations</option>
                <option value="arithmetic-kmers">Arithmetic operations on kmer 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 kmer 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>

                    </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 kmers that occur N times in the input</option>
                        <option value="frequency">Return kmers 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 kmers 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 kmers 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: subsctract x from the count of each k-mer</option>

                <param name="input_meryldb_03" type="data" 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" multiple="true" label="F1 reads" help="Select F1 reads used for generating the assembly" />
                <param name="paternal_reads" type="data" format="fastq,fasta" 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" multiple="true" optional="true" label="Maternal reads" help="Select the maternal reads used for generating the assembly" />
                <conditional name="options_kmer_size">

            </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">

        <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="300" />
                </assert_contents>

        <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="300" />
                </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="300" />
                </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="300"/>
                </assert_contents>

        <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="59378" delta="300" />
                </assert_contents>

                <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>
    </tests>
    <help><![CDATA[

.. class:: infomark

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: 

    ::

        - 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>`_