Merge pull request #6798 from bernt-matthias/umi-fixes

umi-tools fixes
galaxyproject · Feb 28, 2025 · 11a7415 · 11a7415
2 parents 06d653e + be8c3fe
commit 11a7415
Show file tree

Hide file tree

Showing 12 changed files with 30 additions and 37 deletions.
diff --git a/tools/umi_tools/macros.xml b/tools/umi_tools/macros.xml
@@ -9,7 +9,8 @@
     <xml name="requirements">
         <requirements>
             <requirement type="package" version="@TOOL_VERSION@">umi_tools</requirement>
-            <yield />
+            <requirement type="package" version="1.21">samtools</requirement>
+            <requirement type="package" version="4.7">sed</requirement>
         </requirements>
     </xml>
     <xml name="citations">
@@ -551,7 +552,9 @@ count for the gene.
     <token name="@GROUPDEDUP_OPTIONS@"><![CDATA[
         $gd.buffer_whole_contig
         $gd.whole_contig
-        $gd.multimapping_detection_method
+        #if $gd.multimapping_detection_method
+            --multimapping-detection-method $gd.multimapping_detection_method
+        #end if
     ]]></token>
 
     <xml name="log_input_macro">

diff --git a/tools/umi_tools/test-data/dedup_out1.bam b/tools/umi_tools/test-data/dedup_out1.bam
diff --git a/tools/umi_tools/test-data/dedup_out2.bam b/tools/umi_tools/test-data/dedup_out2.bam
diff --git a/tools/umi_tools/test-data/dedup_out3.bam b/tools/umi_tools/test-data/dedup_out3.bam
diff --git a/tools/umi_tools/test-data/dedup_out4.bam b/tools/umi_tools/test-data/dedup_out4.bam
diff --git a/tools/umi_tools/test-data/dedup_out5.bam b/tools/umi_tools/test-data/dedup_out5.bam
diff --git a/tools/umi_tools/test-data/dedup_out6.bam b/tools/umi_tools/test-data/dedup_out6.bam
diff --git a/tools/umi_tools/umi-tools_counts.xml b/tools/umi_tools/umi-tools_counts.xml
@@ -4,11 +4,7 @@
         <import>macros.xml</import>
     </macros>
     <expand macro="bio_tools"/>
-    <expand macro="requirements">
-        <!-- TODO see comment in LINK_SAM_BAM_INPUT -->
-        <requirement type="package" version="1.12">samtools</requirement>
-        <requirement type="package" version="4.7">sed</requirement>
-    </expand>
+    <expand macro="requirements"/>
     <command detect_errors="exit_code"><![CDATA[
         #import re
         @LINK_SAM_BAM_INPUT@        

diff --git a/tools/umi_tools/umi-tools_dedup.xml b/tools/umi_tools/umi-tools_dedup.xml
@@ -4,9 +4,7 @@
         <import>macros.xml</import>
     </macros>
     <expand macro="bio_tools"/>
-    <expand macro="requirements">
-        <requirement type="package" version="1.21">samtools</requirement>
-    </expand>
+    <expand macro="requirements"/>
     <command detect_errors="exit_code"><![CDATA[
         @LINK_SAM_BAM_INPUT@
 
@@ -111,12 +109,12 @@
             <output name="output" file="dedup_out4.bam" ftype="bam" lines_diff="2"/>
         </test>
         <test expect_num_outputs="1">
-            <param name="input" value="group_in5.bam" ftype="bam" />
+            <param name="input" value="group_in4.bam" ftype="bam" />
             <section name="advanced">
                 <param name="random_seed" value="0" />
             </section>
             <conditional name="bc">
-                <param name="extract_umi_method" value="read_id" />
+                <param name="extract_umi_method" value="tag" />
                 <param name="umi_tag" value="BX" />
             </conditional>
             <section name="umi">
@@ -125,12 +123,12 @@
             <output name="output" file="dedup_out5.bam" ftype="bam" lines_diff="2"/>
         </test>
         <test expect_num_outputs="1">
-            <param name="input" value="group_in6.bam" ftype="bam" />
+            <param name="input" value="group_in4.bam" ftype="bam" />
             <section name="advanced">
                 <param name="random_seed" value="0" />
             </section>
             <conditional name="bc">
-                <param name="extract_umi_method" value="read_id" />
+                <param name="extract_umi_method" value="tag" />
                 <param name="umi_tag" value="BX" />
             </conditional>
             <section name="umi">
@@ -139,12 +137,12 @@
             <output name="output" file="dedup_out6.bam" ftype="bam" lines_diff="2"/>
         </test>
         <test expect_num_outputs="5">
-            <param name="input" value="group_in6.bam" ftype="bam" />
+            <param name="input" value="group_in4.bam" ftype="bam" />
             <section name="advanced">
                 <param name="random_seed" value="0" />
             </section>
             <conditional name="bc">
-                <param name="extract_umi_method" value="read_id" />
+                <param name="extract_umi_method" value="tag" />
                 <param name="umi_tag" value="BX" />
             </conditional>
             <section name="umi">

diff --git a/tools/umi_tools/umi-tools_extract.xml b/tools/umi_tools/umi-tools_extract.xml
@@ -83,7 +83,7 @@
     ]]></command>
     <inputs>
         <expand macro="input_types">
-                <param argument="--reconcile-pairs" type="boolean" truevalue="--reconcile-pairs" falsevalue="" checked="false" label="Allow unpaired reads" help="Allow the presences of reads in read2 input that are not present in read1 input. This allows cell barcode filtering of read1s without considering read2s" />
+            <param argument="--reconcile-pairs" type="boolean" truevalue="--reconcile-pairs" falsevalue="" checked="false" label="Allow unpaired reads" help="Allow the presences of reads in read2 input that are not present in read1 input. This allows cell barcode filtering of read1s without considering read2s" />
         </expand>
         <expand macro="fastq_barcode_extraction_options_macro"/>
 
@@ -134,8 +134,8 @@
             <conditional name="extract_method_cond">
                 <param name="prime3" value="true" />
             </conditional>
-            <param name="quality_selector" value="true" />
-            <param name="quality_filter_threshold" value="10" />
+            <param name="quality|quality_selector" value="true" />
+            <param name="quality|quality_filter_threshold" value="10" />
             <param name="log" value="true"/>
             <output name="out" file="out_SE.fastq" ftype="fastqsanger" />
             <output name="out_log" >
@@ -166,7 +166,6 @@
         <test expect_num_outputs="4">
             <conditional name="input_type_cond">
                 <param name="input_type" value="paired_collection" /> <!-- same as before, but uncompressed -->
-                <param name="paired_type" value="no" />
                 <param name="input_readpair">
                     <collection type="paired" >
                         <element name="forward" ftype="fastqsanger" value="t_R1.fastq" />
@@ -194,7 +193,7 @@
                 <param name="input_read2" value="scrb_seq_fastq.2.gz" ftype="fastqsanger.gz" />
                 <param name="bc_pattern" value="CCCCCCNNNNNNNNNN" />
             </conditional>
-            <param name="extract_method" value="string" />
+            <param name="extract_method_cond|extract_method" value="string" />
             <param name="whitelist" value="scrb_seq_barcodes" />
             <param name="log" value="true"/>
             <output name="out2" file="scrb_extract.fastq.gz" decompress="true"  ftype="fastqsanger.gz" />
@@ -206,7 +205,7 @@
                 <param name="input_read2" value="scrb_seq_fastq.2.gz" ftype="fastqsanger.gz" />
                 <param name="bc_pattern" value="^(?P&lt;cell_1&gt;.{6})(?P&lt;umi_1&gt;.{10})" />
             </conditional>
-            <param name="extract_method" value="regex" />
+            <param name="extract_method_cond|extract_method" value="regex" />
             <param name="whitelist" value="scrb_seq_barcodes" />
             <param name="log" value="true"/>
             <output name="out2" file="scrb_extract.fastq.gz" decompress="true" ftype="fastqsanger.gz" />
@@ -218,7 +217,7 @@
                 <param name="input_read2" value="read_R2.200.gz" ftype="fastqsanger.gz" />
                 <param name="bc_pattern" value="NNNNNNCCCCCC" />
             </conditional>
-            <param name="extract_method" value="string" />
+            <param name="extract_method_cond|extract_method" value="string" />
             <output name="out" file="read_R1.200_extracted.fastq.gz" ftype="fastqsanger.gz" decompress="true" lines_diff="1" />
             <output name="out2" file="read_R2.200_extracted.fastq.gz" ftype="fastqsanger.gz" decompress="true" lines_diff="1" />
         </test>

diff --git a/tools/umi_tools/umi-tools_group.xml b/tools/umi_tools/umi-tools_group.xml
@@ -4,9 +4,7 @@
         <import>macros.xml</import>
     </macros>
     <expand macro="bio_tools"/>
-    <expand macro="requirements">
-        <requirement type="package" version="1.21">samtools</requirement>
-    </expand>
+    <expand macro="requirements"/>
     <command detect_errors="exit_code"><![CDATA[
         @LINK_SAM_BAM_INPUT@
 

diff --git a/tools/umi_tools/umi-tools_whitelist.xml b/tools/umi_tools/umi-tools_whitelist.xml
@@ -141,10 +141,10 @@
             <conditional name="extract_method_cond">
                 <param name="prime3" value="false" />
             </conditional>
-            <param name="use_cell_opts" value="advanced" />
+            <param name="celloptions|use_cell_opts" value="advanced" />
+            <param name="celloptions|expect_cells" value="5" />
+            <param name="celloptions|error_correct_threshold" value="3" />
             <param name="knee_method" value="density"/>
-            <param name="expect_cells" value="5" />
-            <param name="error_correct_threshold" value="3" />
             <param name="log" value="true" />
             <output name="out_whitelist" file="out_wl_paired.txt" />
             <output name="out_log" file="out_wl_paired.log" lines_diff="50" />
@@ -167,12 +167,11 @@
             <conditional name="extract_method_cond">
                 <param name="prime3" value="false" />
             </conditional>
-            <param name="use_cell_opts" value="advanced" />
             <param name="knee_method" value="density"/>
-            <param name="expect_cells" value="5" />
-            <param name="error_correct_threshold" value="3" />
+            <param name="celloptions|use_cell_opts" value="advanced" />
+            <param name="celloptions|expect_cells" value="5" />
+            <param name="celloptions|error_correct_threshold" value="3" />
             <param name="log" value="true" />
-            <param name="filtered_out_bool" value="true"/>
             <output name="out_whitelist" file="out_wl_paired.txt" />
             <output name="out_log" file="out_wl_paired.log" lines_diff="50" />
             <output name="out_html_report" file="out_wl_paired.html" />
@@ -183,13 +182,13 @@
             <conditional name="input_type_cond" >
                 <param name="input_type" value="single" />
                 <param name="input_read1" value="testYYY.40k.fastq.gz" ftype="fastqsanger.gz" />
+                <param name="bc_pattern" value="(?P&#60;cell_1&#62;.{8,10})(?P&#60;discard_1&#62;ACTGGCCTGCGA){s&#60;=3}(?P&#60;cell_2&#62;.{9})(?P&#60;discard_2&#62;GGTAGCGGTGACA){s&#60;=3}(?P&#60;cell_3&#62;.{9})(?P&#60;umi_1&#62;.{8})T{3}.*" />
             </conditional>
-            <param name="bc_pattern" value="(?P&#60;cell_1&#62;.{8,10})(?P&#60;discard_1&#62;ACTGGCCTGCGA){s&#60;=3}(?P&#60;cell_2&#62;.{9})(?P&#60;discard_2&#62;GGTAGCGGTGACA){s&#60;=3}(?P&#60;cell_3&#62;.{9})(?P&#60;umi_1&#62;.{8})T{3}.*" />
-            <param name="extract_method" value="regex" />
+            <param name="extract_method_cond|extract_method" value="regex" />
+            <param name="extract_method_cond|filtered_out_bool" value="true"/>
             <param name="method" value="umis" />
             <param name="knee_method" value="density"/>
-            <param name="prime3" value="true" />
-            <param name="filtered_out_bool" value="true"/>
+
             <output name="out_whitelist" file="out_wl_user.single.txt" />
             <output name="out_thresh" file="out_wl_user.single.tresh.tab" />
             <output name="out_html_report" file="out_wl_user.single.html" />