annotate kmc_tools_simple.xml @ 2:c97f8a687258 draft

planemo upload for repository https://github.com/usegalaxy-au/tools-au/tree/master/tools/kmc commit 3be367228b531c346c10700f07d57ae44394be36-dirty

<tool id="kmc_simple" name="KMC simple" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
    <description>simple operations for two input kmer sets</description>
    <macros>
	  <import>macros.xml</import>
    </macros>
    <xrefs>
        <xref type='bio.tools'>kmc</xref>
    </xrefs>
    <expand macro="requirements" />
    <expand macro="stdio" />
    <expand macro="version_command" /> 
    <command><![CDATA[
	    mkdir outdir &&
	    #for $f in $input_collection_one:
	    	#if '.kmc_suf' in str($f.element_identifier)
	    	    ln -s '${f}' kmers_db_one.kmc_suf &&
	        #else
	    	    ln -s '${f}' kmers_db_one.kmc_pre &&
	        #end if
	    #end for
	    #for $f in $input_collection_two:
                #if '.kmc_suf' in str($f.element_identifier)
                    ln -s '${f}' kmers_db_two.kmc_suf &&
                #else
                    ln -s '${f}' kmers_db_two.kmc_pre &&
                #end if
            #end for
	    kmc_tools 
	      -t\${GALAXY_SLOTS:-4}
	      simple
	      #if $input_options_cond.input_param == "None":
	           kmers_db_one
	    	   kmers_db_two
	      #elif $input_options_cond.input_param == "kmc_one_params":
	           kmers_db_one
	     	   -ci'$input_options_cond.kmc_db_one_param.min_kmer_occurrence'
	           -cx'$input_options_cond.kmc_db_one_param.max_kmer_occurrence'
		   kmers_db_two
	      #elif $input_options_cond.input_param == "kmc_two_params":
	           kmers_db_one 
	  	   kmers_db_two
	           -ci'$input_options_cond.kmc_db_two_param.min_kmer_occurrence'
                   -cx'$input_options_cond.kmc_db_two_param.max_kmer_occurrence'
	      #elif $input_options_cond.input_param == "both_kmc_params":
		   kmers_db_one
	           -ci'$input_options_cond.both_db_one_param.min_kmer_occurrence'
                   -cx'$input_options_cond.both_db_one_param.max_kmer_occurrence'
                   kmers_db_two
                   -ci'$input_options_cond.both_db_two_param.min_kmer_occurrence'
                   -cx'$input_options_cond.both_db_two_param.max_kmer_occurrence'
	      #end if
	      #set $operation_arr=[]
	      #for $i,$opt in enumerate($operations)
	           #silent $operation_arr.append(str($opt.ops.select_operation))
	    	   #if str($opt.ops.select_operation) == "intersect":
	    		$opt.ops.select_operation
			#if str($opt.ops.add_output_params.oc) != 'None':  
	    		    '${opt.ops.select_operation}_${opt.ops.add_output_params.oc}_db'
	                     -oc'$opt.ops.add_output_params.oc'
	    		#else
	    		    '${opt.ops.select_operation}_min_db'
	    		    -oc'min'
	    		#end if
		   	#if str($opt.ops.add_output_params.min_kmer_occurrence) != '':
	           		-ci'$opt.ops.add_output_params.min_kmer_occurrence'
	           	#end if
	           	#if str($opt.ops.add_output_params.max_kmer_occurrence) != '':
	           		-cx'$opt.ops.add_output_params.max_kmer_occurrence'
	    	   	#end if
	    	   #elif str($opt.ops.select_operation) == "union":
			$opt.ops.select_operation
                        #if str($opt.ops.add_output_params.oc) != 'None':
                            '${opt.ops.select_operation}_${opt.ops.add_output_params.oc}_db'
                            -oc'$opt.ops.add_output_params.oc'
                        #else  
	    	            '${opt.ops.select_operation}_sum_db' 
	    		     -oc'sum'
                        #end if
                        #if str($opt.ops.add_output_params.min_kmer_occurrence) != '':
                                -ci'$opt.ops.add_output_params.min_kmer_occurrence'
                        #end if
                        #if str($opt.ops.add_output_params.max_kmer_occurrence) != '':
                                -cx'$opt.ops.add_output_params.max_kmer_occurrence'
	    		#end if
 		   #elif str($opt.ops.select_operation) == "kmers_subtract":
                        $opt.ops.select_operation
	    		${opt.ops.select_operation}_db
	    		#if str($opt.ops.add_output_params.min_kmer_occurrence) != '':
                                -ci'$opt.ops.add_output_params.min_kmer_occurrence'
                        #end if
                        #if str($opt.ops.add_output_params.max_kmer_occurrence) != '':
                                -cx'$opt.ops.add_output_params.max_kmer_occurrence'
	    	        #end if
		   #elif str($opt.ops.select_operation) == "reverse_kmers_subtract":
                        $opt.ops.select_operation
                        ${opt.ops.select_operation}_db
                        #if str($opt.ops.add_output_params.min_kmer_occurrence) != '':
                                -ci'$opt.ops.add_output_params.min_kmer_occurrence'
                        #end if
                        #if str($opt.ops.add_output_params.max_kmer_occurrence) != '':
                                -cx'$opt.ops.add_output_params.max_kmer_occurrence'
	    		#end if
                   #elif str($opt.ops.select_operation) == "counters_subtract":
                        $opt.ops.select_operation
                        #if str($opt.ops.add_output_params.oc) != 'None':
                            '${opt.ops.select_operation}_${opt.ops.add_output_params.oc}_db'
                            -oc'$opt.ops.add_output_params.oc'
                        #else
                            '${opt.ops.select_operation}_min_db'
                            -oc'diff'
                        #end if
                        #if str($opt.ops.add_output_params.min_kmer_occurrence) != '':
                                -ci'$opt.ops.add_output_params.min_kmer_occurrence'
                        #end if
                        #if str($opt.ops.add_output_params.max_kmer_occurrence) != '':
                                -cx'$opt.ops.add_output_params.max_kmer_occurrence'
                        #end if
                   #elif str($opt.ops.select_operation) == "reverse_counters_subtract":
                        $opt.ops.select_operation
                        #if str($opt.ops.add_output_params.oc) != 'None':
                            '${opt.ops.select_operation}_${opt.ops.add_output_params.oc}_db'
                            -oc'$opt.ops.add_output_params.oc'
                        #else
                            '${opt.ops.select_operation}_min_db'
                            -oc'diff'
                        #end if
                        #if str($opt.ops.add_output_params.min_kmer_occurrence) != '':
                                -ci'$opt.ops.add_output_params.min_kmer_occurrence'
                        #end if
                        #if str($opt.ops.add_output_params.max_kmer_occurrence) != '':
                                -cx'$opt.ops.add_output_params.max_kmer_occurrence'
                        #end if
	           #end if 
	     #end for
	    
	     #for $i,$op in enumerate($operation_arr):
	        #if str($op) == "kmers_subtract" or str($op) == "reverse_kmers_subtract":
	         && cp '${op}'_db.kmc_suf outdir/
	         && cp '${op}'_db.kmc_pre outdir/
                 2>&1 | tee -a '$out_log'
	       #else
		 && cp '${op}'_*_db.kmc_suf outdir/
	         && cp '${op}'_*_db.kmc_pre outdir/
                 2>&1 | tee -a '$out_log'
	       #end if
	     #end for
	    ]]></command>
    <inputs>
	    <param name="input_collection_one" type="data_collection" collection_type="list" label="KMC db one"/>
	    <param name="input_collection_two" type="data_collection" collection_type="list" label="KMC db two"/>
	    <conditional name="input_options_cond">
		    <param name="input_param" type="select" label="Input parameters">
			 <option value="None" selected="true">Both KMC DB with default parameters</option>
			 <option value="kmc_one_params">KMC db one only</option>
			 <option value="kmc_two_params">KMC db two only</option>
			 <option value="both_kmc_params">Both KMC DB with user parameters</option>
		    </param>
		    <when value="None"/>
		    <when value="kmc_one_params">
			 <section name="kmc_db_one_param" title="KMC DB one parameters">
				 <expand macro="input_option"/>
			 </section>
		    </when>
		    <when value="kmc_two_params">
			 <section name="kmc_db_two_param" title="KMC DB two parameters">
				 <expand macro="input_option"/>
			 </section>
		    </when>
		    <when value="both_kmc_params">
			 <section name="both_db_one_param" title="KMC DB one parameters">
				 <expand macro="input_option"/>
			 </section>
			 <section name="both_db_two_param" title="KMC DB two parameters">
				 <expand macro="input_option"/>
			 </section>
		    </when>
	    </conditional>
	    <repeat name="operations" title="Operations" min="1" max="3">
		   <conditional name="ops">
		     <expand macro="simple_operation"/>
		     <when value="intersect">
			    <section name="add_output_params" expanded="false" title="Additional output parameters">
                               <expand macro="counter_calculation_option"/>
                            </section>
                     </when>
		     <when value="union">
			    <section name="add_output_params" expanded="false" title="Additional output parameters">
                               <expand macro="counter_calculation_option"/>
                            </section>
		    </when>
		    <when value="kmers_subtract">
			    <section name="add_output_params" expanded="false" title="Additional output parameters">
                               <expand macro="none_counter_calculation_option"/>
                            </section>
		    </when>
                    <when value="reverse_kmers_subtract">
                            <section name="add_output_params" expanded="false" title="Additional output parameters">
                               <expand macro="none_counter_calculation_option"/>
                            </section>
		    </when>
                    <when value="counters_subtract">
                            <section name="add_output_params" expanded="false" title="Additional output parameters">
                               <expand macro="counter_calculation_option"/>
                            </section>
                    </when>
                    <when value="reverse_counters_subtract">
                            <section name="add_output_params" expanded="false" title="Additional output parameters">
                               <expand macro="counter_calculation_option"/>
                            </section>
                    </when>
		  </conditional>
          </repeat>
    </inputs>
    <outputs>
          <collection name="kmc_db" type="list" label="${tool.name} on ${on_string}: kmc db">
		  <discover_datasets pattern="(?P&lt;designation&gt;.+)" format="binary" directory="outdir/" />
	  </collection>
          <data name="out_log" format="txt" label="${tool.name} on ${on_string}: log"/>
    </outputs>
    <tests>
	 <test>
	    <!-- #1 test with default parameters -->
            <param name="input_collection_one">
                <collection type="list">
		    <element name="kmers_one.kmc_pre" value="filtered_kmers1.kmc_pre"/>
		    <element name="kmers_one.kmc_suf" value="filtered_kmers1.kmc_suf"/>
                </collection>
	    </param>
            <param name="input_collection_two">
                <collection type="list">
		    <element name="kmers_two.kmc_pre" value="filtered_kmers2.kmc_pre"/>
		    <element name="kmers_two.kmc_suf" value="filtered_kmers2.kmc_suf"/>
                </collection>
	    </param>
	    <conditional name="input_options_cond">
		   <param name="input_param" value="None"/>
	    </conditional>
	    <param name="select_operation" value="intersect" />
	    <section name="add_output_params">
		   <param name="oc" value="max"/>
	    </section>
            <output_collection name="kmc_db" type="list">
                <element name="intersect_max_db.kmc_pre" file="intersect_max_db.kmc_pre"/>
                <element name="intersect_max_db.kmc_suf" file="intersect_max_db.kmc_suf"/>
            </output_collection>
         </test>
         <test>
            <!-- #2 test with input parameters for KMC DB one only -->
            <param name="input_collection_one">
                <collection type="list">
                    <element name="kmers_one.kmc_pre" value="filtered_kmers1.kmc_pre"/>
                    <element name="kmers_one.kmc_suf" value="filtered_kmers1.kmc_suf"/>
                </collection>
            </param>
            <param name="input_collection_two">
                <collection type="list">
                    <element name="kmers_two.kmc_pre" value="filtered_kmers2.kmc_pre"/>
                    <element name="kmers_two.kmc_suf" value="filtered_kmers2.kmc_suf"/>
                </collection>
            </param>
            <conditional name="input_options_cond">
                   <param name="input_param" value="kmc_one_params"/>
            </conditional>
	    <param name="select_operation" value="intersect" />
	    <section name="kmc_db_one_param">
		   <param name="min_kmer_occurrence" value="3"/>
		   <param name="max_kmer_occurrence" value="30"/>
	    </section>
            <section name="add_output_params">
                   <param name="oc" value="min"/>
            </section>
            <output_collection name="kmc_db" type="list">
                <element name="intersect_min_db.kmc_pre" file="intersect_min_db.kmc_pre"/>
                <element name="intersect_min_db.kmc_suf" file="intersect_min_db.kmc_suf"/>
            </output_collection>
         </test>
         <test>
            <!-- #3 test with input parameters for KMC DB two only -->
            <param name="input_collection_one">
                <collection type="list">
                    <element name="kmers_one.kmc_pre" value="filtered_kmers1.kmc_pre"/>
                    <element name="kmers_one.kmc_suf" value="filtered_kmers1.kmc_suf"/>
                </collection>
            </param>
            <param name="input_collection_two">
                <collection type="list">
                    <element name="kmers_two.kmc_pre" value="filtered_kmers2.kmc_pre"/>
                    <element name="kmers_two.kmc_suf" value="filtered_kmers2.kmc_pre"/>
                </collection>
            </param>
            <conditional name="input_options_cond">
                   <param name="input_param" value="kmc_two_params"/>
            </conditional>
            <param name="select_operation" value="union" />
            <section name="kmc_db_two_param">
                   <param name="min_kmer_occurrence" value="3"/>
                   <param name="max_kmer_occurrence" value="30"/>
	    </section>
            <section name="add_output_params">
                   <param name="oc" value="sum"/>
            </section>
            <output_collection name="kmc_db" type="list">
                <element name="union_sum_db.kmc_pre" file="union_sum_db.kmc_pre"/>
                <element name="union_sum_db.kmc_suf" file="union_sum_db.kmc_suf"/>
            </output_collection>
         </test>
         <test>
            <!-- #4 test with input parameters for both KMC DB -->
            <param name="input_collection_one">
                <collection type="list">
                    <element name="kmers_one.kmc_pre" value="filtered_kmers1.kmc_pre"/>
                    <element name="kmers_one.kmc_suf" value="filtered_kmers1.kmc_pre"/>
                </collection>
            </param>
            <param name="input_collection_two">
                <collection type="list">
                    <element name="kmers_two.kmc_pre" value="filtered_kmers2.kmc_pre"/>
                    <element name="kmers_two.kmc_suf" value="filtered_kmers2.kmc_suf"/>
                </collection>
            </param>
            <conditional name="input_options_cond">
                   <param name="input_param" value="both_kmc_params"/>
            </conditional>
            <param name="select_operation" value="counters_subtract" />
            <section name="both_db_one_param">
                   <param name="min_kmer_occurrence" value="1"/>
                   <param name="max_kmer_occurrence" value="10"/>
	    </section>
            <section name="both_db_two_param">
                   <param name="min_kmer_occurrence" value="2"/>
                   <param name="max_kmer_occurrence" value="20"/>
	    </section>
            <section name="add_output_params">
                   <param name="oc" value="sum"/>
            </section>
            <output_collection name="kmc_db" type="list">
                <element name="counters_subtract_sum_db.kmc_pre" file="counters_subtract_sum_db.kmc_pre"/>
                <element name="counters_subtract_sum_db.kmc_suf" file="counters_subtract_sum_db.kmc_suf"/>
            </output_collection>
         </test>
    </tests>
    <help><![CDATA[

.. class:: infomark


**What it does**
	    
*KMC simple operation performs set operation on two input KMC's databases*

**General Syntax**

kmc_tools simple <input1 [input1_params]> <input2 [input2_params]> <oper1 output1 [output_params1]> [<oper2 output2 [output_params2]> ...]

**Input:**
	    
- input file - input1 and input2 are the databases generated by KMC ( kmc file suffix - kmc_pre and kmc_suf )

**Available Operations:**

- intersect - output database will contains only k-mers that are present in both input sets
- union - output database will contains each k-mer present in any of input sets
- kmers_subtract - difference of input sets based on k-mers. Output database will contains only k-mers that are present in first input set but absent in the second one
- counters_subtract - difference of input sets based on k-mers and their counters (weaker version of kmers_subtract). Output database will contains all k-mers that are present in first input, beyond those for which counter operation will lead to remove (i.e. counter equal to 0 or negative number)
- reverse_kmers_subtract - same as kmers_subtract but treat input2 as first and input1 as second
- reverse_counters_subtract - same as counters_subtract but treat input2 as first and input1 as second

**Additional Parameters for inputs:**

- ci<value> - exclude k-mers occurring less than <value> times		
- cx<value> - exclude k-mers occurring more of than <value> times

**Additional Parameters for outputs:**

- ci<value>  - exclude k-mers occurring less than <value> times
- cx<value>  - exclude k-mers occurring more of than <value> times
- cs<value>  - maximal value of a counter
- o<kmc|kff> - output in KMC or KFF format (default: kmc)
- oc<value>  - redefine counter calculation mode for equal k-mers

**Available oc<value>**

- min   - get lower value of a k-mer counter (default value for intersect operation)
- max   - get upper value of a k-mer counter
- sum   - get sum of counters from both databases
- diff  - get difference between counters (default for counters_subtract operation)
- left  - get counter from first database (input1)
- right - get counter from second database (input2)


**Example(1):**

- kmc -k27 file1.fastq kmers1 kmc_tmp_dir
- kmc -k27 file2.fastq kmers2 kmc_tmp_dir
- kmc_tools simple kmers1 -ci10 -cx200 kmers2 -ci4 -cx100 intersect kmers1_kmers2_intersect -ci20 -cx150

**Output(1)**

- kmers1_kmers2_intersect.kmc_suf (i.e intersect_kmc_suf in galaxy)
- kmers1_kmers2_intersect.kmc_pre (i.e intersect_kmc_pre in galaxy)

**Example(2):**

- kmc -k27 file1.fastq kmers1 kmc_tmp_dir
- kmc -k27 file2.fastq kmers2 kmc_tmp_dir
- kmc_tools simple kmers1 kmers2 intersect inter_k1_k2_max -ocmax intersect inter_k1_k2_min union union_k1_k2 -ci10

**Output(2)**

- inter_k1_k2_max.kmc_suf (i.e intersect.kmc_suf in galaxy)
- inter_k1_k2_max.kmc_pre (i.e intersect.kmc_pre in galaxy)
- inter_k1_k2_min.kmc_suf (i.e intersect.kmc_suf in galaxy)
- inter_k1_k2_min.kmc_pre (i.e intersect.kmc_pre in galaxy)
- union_k1_k2.kmc_suf (i.e union.kmc_suf in galaxy)
- union_k1_k2.kmc_pre (i.e union.kmc_pre in galaxy
 
.. class:: infomark
	    
**References**

More information are available on `website <https://github.com/refresh-bio/KMC/blob/master/kmc_tools.pdf>`_.
	]]></help>
     <citations>
        <citation type="doi">DOI: 10.1093/bioinformatics/btx304</citation>
	<citation type="doi">DOI: 10.1093/bioinformatics/btv022</citation>
	<citation type="doi">DOI: 10.1186/1471-2105-14-160</citation>
    </citations>
</tool>