--- a/polypolish.xml	Thu Sep 22 07:50:58 2022 +0000
+++ b/polypolish.xml	Tue Nov 12 09:56:19 2024 +0000
@@ -5,45 +5,51 @@
     <macros>
         <import>macro.xml</import>
     </macros>
-    <expand macro='xrefs'/>
-    <expand macro="requirements" />
-    <expand macro="version_command" />
+    <expand macro="xrefs"/>
+    <expand macro="requirements"/>
+    <expand macro="version_command"/>
     <command detect_errors="aggressive"><![CDATA[
         ln -s '$input.fasta_file' input_data &&
         #*======================================
                     For single SAM
         ======================================*#
         #if $input.sam_data_type.sam_selector == 'single'
-            #if $input.sam_data_type.single_sam.ext == 'unsorted.bam'
-                samtools view -h $input.sam_data_type.single_sam > input_sam &&
+            #if $input.sam_data_type.single_sam.is_of_type("unsorted.bam")
+                samtools view -@ \$addthreads -h $input.sam_data_type.single_sam > input_sam &&
             #elif $input.sam_data_type.single_sam.ext == 'sam'
                 ln -s $input.sam_data_type.single_sam input_sam &&
+            #else
+                echo "${input.sam_data_type.single_sam} not a sam/bam file"
             #end if
-            polypolish input_data input_sam > '$polished_fasta'
+            polypolish polish input_data input_sam > '$polished_fasta'
         #*======================================
                     For paired SAM
         ======================================*#
         #elif $input.sam_data_type.sam_selector == 'paired'
-            #if $input.sam_data_type.R1_sam.ext == 'unsorted.bam'
-                samtools view -h $input.sam_data_type.R1_sam > sample_R1.sam &&
+            #if $input.sam_data_type.R1_sam.is_of_type("unsorted.bam")
+                samtools view -@ \$addthreads -h $input.sam_data_type.R1_sam > sample_R1.sam &&
             #elif $input.sam_data_type.R1_sam.ext == 'sam'
                 ln -s '$input.sam_data_type.R1_sam' sample_R1.sam &&
+            #else
+                echo "${input.sam_data_type.single_sam} not a sam/bam file"
             #end if
-            #if $input.sam_data_type.R2_sam.ext == 'unsorted.bam'
-                samtools view -h $input.sam_data_type.R2_sam > sample_R2.sam &&
+            #if $input.sam_data_type.R2_sam.is_of_type("unsorted.bam")
+                samtools view -@ \$addthreads -h $input.sam_data_type.R2_sam > sample_R2.sam &&
             #elif $input.sam_data_type.R2_sam.ext == 'sam'
                 ln -s '$input.sam_data_type.R2_sam' sample_R2.sam &&
+            #else
+                echo "${input.sam_data_type.single_sam} not a sam/bam file"
             #end if
             #if $input.sam_data_type.insert_filter.filter_select == 'filter'
-                polypolish_insert_filter.py --low '$input.sam_data_type.insert_filter.low'
+                polypolish filter --low '$input.sam_data_type.insert_filter.low'
                                             --high '$input.sam_data_type.insert_filter.high'
                                             --in1 sample_R1.sam
                                             --in2 sample_R2.sam
                                             --out1 'filtered_1.sam'
                                             --out2 'filtered_2.sam' &&
-                polypolish input_data 'filtered_1.sam' 'filtered_2.sam' >  $polished_fasta
+                polypolish polish input_data 'filtered_1.sam' 'filtered_2.sam' >  $polished_fasta
             #else
-                polypolish input_data sample_R1.sam sample_R2.sam  >  $polished_fasta
+                polypolish polish input_data sample_R1.sam sample_R2.sam  >  $polished_fasta
             #end if
         #*======================================
             For multiple single-end SAM
@@ -51,31 +57,33 @@
         #elif $input.sam_data_type.sam_selector == 'multiple_single'
             mkdir single_collection &&
             #for $value, $single_sam in enumerate($input.sam_data_type.single_collection):
-                #if $single_sam.ext == 'unsorted.bam'
-                    samtools view -h $single_sam > 'single_collection/$(single_sam.element_identifier).sam' &&
+                #if $single_sam.is_of_type("unsorted.bam")
+                    samtools view -@ \$addthreads -h $single_sam > 'single_collection/$(single_sam.element_identifier).sam' &&
                 #elif $single_sam.ext == 'sam'
                     ln -s $single_sam 'single_collection/$(single_sam.element_identifier).$(single_sam.ext)' &&
+                #else
+                    echo "${single_sam} is not a sam/bam file"
                 #end if
             #end for
-            polypolish input_data single_collection/*.sam > '$polished_fasta'
+            polypolish polish input_data single_collection/*.sam > '$polished_fasta'
         #*======================================
             For multiple paired-end SAM
         ======================================*#
         #elif $input.sam_data_type.sam_selector == "multiple_paired"
             mkdir paired_collection &&
             #for $value, $paired_sam in enumerate($input.sam_data_type.paired_collection):
-                #if $paired_sam.forward.ext == 'unsorted.bam'
-                    samtools view -h $paired_sam.forward > 'paired_collection/forward_input$(value)$(paired_sam.forward.element_identifier).sam' &&
+                #if $paired_sam.forward.is_of_type("unsorted.bam")
+                    samtools view -@ \$addthreads -h $paired_sam.forward > 'paired_collection/forward_input$(value)$(paired_sam.forward.element_identifier).sam' &&
                 #else
                     ln -s '$paired_sam.forward' 'paired_collection/forward_input$(value)$(paired_sam.forward.element_identifier).sam' &&
                 #end if
-                #if $paired_sam.reverse.ext == 'unsorted.bam'
-                    samtools view -h $paired_sam.reverse > 'paired_collection/reverse_input$(value)$(paired_sam.reverse.element_identifier).sam' &&
+                #if $paired_sam.reverse.is_of_type("unsorted.bam")
+                    samtools view -@ \$addthreads -h $paired_sam.reverse > 'paired_collection/reverse_input$(value)$(paired_sam.reverse.element_identifier).sam' &&
                 #else
                     ln -s '$paired_sam.reverse' 'paired_collection/reverse_input$(value)$(paired_sam.reverse.element_identifier).sam' &&
                 #end if
                 #if $input.sam_data_type.insert_filter.filter_select == 'filter'
-                    polypolish_insert_filter.py --low '$input.sam_data_type.insert_filter.low'
+                    polypolish filter --low '$input.sam_data_type.insert_filter.low'
                                                 --high '$input.sam_data_type.insert_filter.high'
                                                 --in1 'paired_collection/forward_input$(value)$(paired_sam.forward.element_identifier).sam'
                                                 --in2 'paired_collection/reverse_input$(value)$(paired_sam.reverse.element_identifier).sam'
@@ -87,9 +95,9 @@
                     Filtering option
         ======================================*#
             #if $input.sam_data_type.insert_filter.filter_select == 'filter'
-                polypolish input_data paired_collection/*_filtered.sam > '$polished_fasta'
+                polypolish polish input_data paired_collection/*_filtered.sam > '$polished_fasta'
             #else
-                polypolish input_data paired_collection/*.sam > '$polished_fasta'
+                polypolish polish input_data paired_collection/*.sam > '$polished_fasta'
             #end if
         #end if
         #*======================================
@@ -108,8 +116,7 @@
     </command>
     <inputs>
         <section name="input" title="Input sequences" expanded="True">
-            <param name="fasta_file" type="data" format="fasta" label="Select a draft genome for polishing"
-                   help="Fasta sequence to be cleaned using short-reads data"/>
+            <param name="fasta_file" type="data" format="fasta" label="Select a draft genome for polishing" help="Fasta sequence to be cleaned using short-reads data"/>
             <conditional name="sam_data_type">
                 <param name="sam_selector" type="select" label="Select aligned data to polish" help="Choose number of aligned sam/bam files. Need aligned file with all possible locations in aligner option">
                     <option value="single">Single SAM/BAM file</option>
@@ -135,16 +142,12 @@
             </conditional>
         </section>
         <section name="options" title="Options" expanded="False">
-            <param name="min_depth" argument="--min_depth" type="integer" min="0" value="5" label="Minimal depth"
-                   help="A base must occur at least this many times in the pileup to be considered valid [default: 5]"/>
-            <param name="fraction_invalid" argument="--fraction_invalid" type="float" min="0" value="0.2" max="1" label="Minimal invalid fraction"
-                   help="A base must make up less than this fraction of the read depth to be considered invalid [default: 0.2]"/>
-            <param name="max_errors" argument="--max_errors" type="integer" min="0" value="10" label="Number of mismatch/indels to ignore alignments"
-                   help="Ignore alignments with more than this many mismatches and indels [default: 10]"/>
-            <param name="fraction_valid" argument="--fraction_valid" type="float" min="0" value="0.5" max="1" label="Minimal valid fraction"
-                   help="A base must make up at least this fraction of the read depth to be considered valid [default: 0.5"/>
+            <param argument="--min_depth" type="integer" min="0" value="5" label="Minimal depth" help="A base must occur at least this many times in the pileup to be considered valid [default: 5]"/>
+            <param argument="--fraction_invalid" type="float" min="0" value="0.2" max="1" label="Minimal invalid fraction" help="A base must make up less than this fraction of the read depth to be considered invalid [default: 0.2]"/>
+            <param argument="--max_errors" type="integer" min="0" value="10" label="Number of mismatch/indels to ignore alignments" help="Ignore alignments with more than this many mismatches and indels [default: 10]"/>
+            <param argument="--fraction_valid" type="float" min="0" value="0.5" max="1" label="Minimal valid fraction" help="A base must make up at least this fraction of the read depth to be considered valid [default: 0.5"/>
             <param name="keep_logfile" type="boolean" truevalue="true" falsevalue="false" label="Keep log file"/>
-            <param name="debug" argument="--debug" type="boolean" truevalue="true" falsevalue="false" label="Keep per base information file"/>
+            <param argument="--debug" type="boolean" truevalue="true" falsevalue="false" label="Keep per base information file"/>
         </section>
     </inputs>
     <outputs>
@@ -343,9 +346,9 @@
                     <param name="sam_selector" value="multiple_single"/>
                     <param name="single_collection">
                         <collection type="list">
-                            <element name="R1_sam" value="aligned_test_file/alignement_R1.bam" ftype="unsorted.bam"/>
-                            <element name="R1_bis_sam" value="aligned_test_file/alignement_R1_bis.bam" ftype="unsorted.bam"/>
-                            <element name="R1_ter_sam" value="aligned_test_file/alignement_R1_ter.bam" ftype="unsorted.bam"/>
+                            <element name="R1_sam" value="aligned_test_file/alignement_R1.sam" ftype="sam"/>
+                            <element name="R1_bis_sam" value="aligned_test_file/alignement_R1_bis.sam" ftype="sam"/>
+                            <element name="R1_ter_sam" value="aligned_test_file/alignement_R1_ter.sam" ftype="sam"/>
                         </collection>
                     </param>
                 </conditional>
@@ -361,29 +364,29 @@
             <section name="input">
                 <param name="fasta_file" value="contigs.fa"/>
                 <conditional name="sam_data_type">
-                  <param name="sam_selector" value="multiple_paired"/>
-                  <param name="paired_collection">
-                      <collection type="list:paired">
-                          <element name="paired_1">
-                              <collection type="paired">
-                                  <element name="forward" value="aligned_test_file/alignement_R1.bam" ftype="unsorted.bam"/>
-                                  <element name="reverse" value="aligned_test_file/alignement_R2.bam" ftype="unsorted.bam"/>
-                              </collection>
-                          </element>
-                          <element name="paired_2">
-                              <collection type="paired">
-                                  <element name="forward" value="aligned_test_file/alignement_R1_bis.bam" ftype="unsorted.bam"/>
-                                  <element name="reverse" value="aligned_test_file/alignement_R2_bis.bam" ftype="unsorted.bam"/>
-                              </collection>
-                          </element>
-                          <element name="paired_3">
-                              <collection type="paired">
-                                  <element name="forward" value="aligned_test_file/alignement_R1_ter.bam" ftype="unsorted.bam"/>
-                                  <element name="reverse" value="aligned_test_file/alignement_R2_ter.bam" ftype="unsorted.bam"/>
-                              </collection>
-                          </element>
-                      </collection>
-                  </param>
+                    <param name="sam_selector" value="multiple_paired"/>
+                    <param name="paired_collection">
+                        <collection type="list:paired">
+                            <element name="paired_1">
+                                <collection type="paired">
+                                    <element name="forward" value="aligned_test_file/alignement_R1.bam" ftype="unsorted.bam"/>
+                                    <element name="reverse" value="aligned_test_file/alignement_R2.bam" ftype="unsorted.bam"/>
+                                </collection>
+                            </element>
+                            <element name="paired_2">
+                                <collection type="paired">
+                                    <element name="forward" value="aligned_test_file/alignement_R1_bis.bam" ftype="unsorted.bam"/>
+                                    <element name="reverse" value="aligned_test_file/alignement_R2_bis.bam" ftype="unsorted.bam"/>
+                                </collection>
+                            </element>
+                            <element name="paired_3">
+                                <collection type="paired">
+                                    <element name="forward" value="aligned_test_file/alignement_R1_ter.bam" ftype="unsorted.bam"/>
+                                    <element name="reverse" value="aligned_test_file/alignement_R2_ter.bam" ftype="unsorted.bam"/>
+                                </collection>
+                            </element>
+                        </collection>
+                    </param>
                 </conditional>
             </section>
             <section name="options">
@@ -392,8 +395,36 @@
             </section>
             <output name="polished_fasta" value="polished.fasta"/>
         </test>
+        <!-- Test_12 Test with bam files -->
+        <test expect_num_outputs="1">
+            <section name="input">
+                <param name="fasta_file" value="contigs.fa"/>
+                <conditional name="sam_data_type">
+                    <param name="sam_selector" value="single"/>
+                    <param name="single_sam" value="aligned_test_file/alignement_R1.bam"/>
+                </conditional>
+            </section>
+            <output name="polished_fasta" value="polished.fasta"/>
+        </test>
+        <!-- Test_13 paired bam -->
+        <test expect_num_outputs="2">
+            <section name="input">
+                <param name="fasta_file" value="contigs.fa"/>
+                <conditional name="sam_data_type">
+                    <param name="sam_selector" value="paired"/>
+                    <param name="R1_sam" value="aligned_test_file/alignement_R1.bam"/>
+                    <param name="R2_sam" value="aligned_test_file/alignement_R2.bam"/>
+                </conditional>
+            </section>
+            <section name="options">
+                <param name="debug" value="true"/>
+                <param name="keep_logfile" value="false"/>
+            </section>
+            <output name="polished_fasta" value="polished.fasta"/>
+            <output name="debug_file" value="debug_file_test_2.tsv"/>
+        </test>
     </tests>
-  <help><![CDATA[
+    <help><![CDATA[
     **What it does**
     Polypolish is a tool for polishing genome assemblies with short reads.
     Polypolish uses SAM/BAM files where each read has been aligned to all possible locations (not just a single best location).
@@ -401,10 +432,12 @@
 
     **Polypolish pipeline steps**
     1. [Optional] Filter aligned reads
-        - Exclude some alignments based on their insert size
-        - This should reduce the number of excessive alignments, particularly near the edges of repeat sequences, improving Polypolish's ability to fix errors in those regions.
+    Exclude some alignments based on their insert size
+    This should reduce the number of excessive alignments,
+    particularly near the edges of repeat sequences, improving Polypolish's ability to fix errors in those regions.
     2. Clean assembly with filtered reads
 
+
     **Inputs**
     Polypolish need SAM/BAM input format obtain from aligner with option to keep all possible location
     Polypolish take on or more assembly as input fasta.
@@ -414,9 +447,7 @@
     For example using bwa mem to align raw data before use, you need :
     1. To align each read data independantly (also for paired data)
     2. Set the option "Output all alignments for single-ends or unpaired paired-ends" in Select analysis mode>Set input/output options
-      - This allow multiple ailgnemnt output need to use polypolish
-
-      
+    This allow multiple ailgnemnt output need to use polypolish
   ]]></help>
     <expand macro="citations"/>
 </tool>