Merge pull request #5127 from wm75/ivar-update

ivar update
galaxyproject · Feb 13, 2023 · 02d1d48 · 02d1d48
2 parents 28a15bb + d5cd4a3
commit 02d1d48
Show file tree

Hide file tree

Showing 8 changed files with 61 additions and 47 deletions.
diff --git a/tools/ivar/ivar_consensus.xml b/tools/ivar/ivar_consensus.xml
@@ -1,4 +1,4 @@
-<tool id="ivar_consensus" name="ivar consensus" version="@TOOL_VERSION@+galaxy2" profile="@PROFILE@">
+<tool id="ivar_consensus" name="ivar consensus" version="@TOOL_VERSION@+galaxy0" profile="@PROFILE@">
     <description>Call consensus from aligned BAM file</description>
     <macros>
         <import>macros.xml</import>
@@ -13,6 +13,7 @@
         -p consensus
         -q $min_qual
         -t $min_freq
+        -c $min_indel_freq
         -m $min_depth
         $depth_action
         &&
@@ -34,6 +35,7 @@
         ]]>
             </help>
         </param>
+        <param name="min_indel_freq" argument="-c" type="float" min="0" max="1" value="0.8" label="Minimum indel frequency threshold" help="Same meaning as the &quot;Minimum frequency threshold&quot; above, but applied to indels."/>
         <param name="min_depth" argument="-m" type="integer" min="1" max="255" value="10" label="Minimum depth to call consensus"/>
         <param name="depth_action" type="select" label="How to represent positions with coverage less than the minimum depth threshold">
             <option value="-k">Drop from output (-k)</option>
@@ -60,12 +62,14 @@
     
     samtools mpileup -A -a -d 0 -Q 0 sorted.bam | ivar consensus [options]
 
-    There are four parameters that can be set:
+    There are five parameters that can be set:
 
     - **Minimum quality**: the minimum quality of a base to be considered in calculations of variant frequencies at a given position
 
     - **Minimum frequency threshold**: the minimum frequency that the most likely base must surpass to be called as the consensus base at a position.
-    
+
+    - **Minimum indel frequency threshold**: has the same meaning as the previous threshold, but gets applied to indels specifically. Setting this threshold higher than the threshold applied to SNVs helps reduce indel artefacts in the generated consensus.
+
     - **Minimum depth to call consensus**: the minimum required depth to call a consensus base
 
     - **How to represent positions with coverage less than the minimum depth threshold**: for positions for which the above minimum depth to call a consensus base is not reached, you can choose one of three different actions:

diff --git a/tools/ivar/ivar_filtervariants.xml b/tools/ivar/ivar_filtervariants.xml
@@ -1,4 +1,4 @@
-<tool id="ivar_filtervariants" name="ivar filtervariants" version="@TOOL_VERSION@+galaxy1" profile="@PROFILE@">
+<tool id="ivar_filtervariants" name="ivar filtervariants" version="@TOOL_VERSION@+galaxy0" profile="@PROFILE@">
     <description>Filter variants across replicates or multiple samples aligned using the same reference</description>
         <macros>
         <import>macros.xml</import>

diff --git a/tools/ivar/ivar_removereads.xml b/tools/ivar/ivar_removereads.xml
@@ -1,4 +1,4 @@
-<tool id="ivar_removereads" name="ivar removereads" version="@TOOL_VERSION@+galaxy4" profile="@PROFILE@">
+<tool id="ivar_removereads" name="ivar removereads" version="@TOOL_VERSION@+galaxy0" profile="@PROFILE@">
     <description>Remove reads from trimmed BAM file</description>
     <macros>
         <import>macros.xml</import>
@@ -60,9 +60,14 @@
                 <param name="computed" value="no"/>
                 <param name="amplicon_info" value="zika/db/pair_information.tsv"/>
             </conditional>
+            <assert_stdout>
+                <!-- check that primer pairs are picked up correctly
+                from the amplicon info file -->
+                <has_text text="400_18_out_L**&#009;400_18_out_R**&#009;400_23_out_L&#009;400_23_out_R&#009;400_32_out_L&#009;400_32_out_R"/>
+            </assert_stdout>
             <output name="output_bam" ftype="bam">
                 <assert_contents>
-                    <has_size value="3185672" delta="1000"/>
+                    <has_size value="3130923" delta="1000"/>
                 </assert_contents>
             </output>
         </test>
@@ -73,9 +78,14 @@
             <conditional name="amplicons">
                 <param name="computed" value="yes"/>
             </conditional>
+            <assert_stdout>
+                <!-- check that primer pairs are picked up correctly
+                from the autogenerated amplicon info file -->
+                <has_text text="400_18_out_L**&#009;400_18_out_R**&#009;400_23_out_L&#009;400_23_out_R&#009;400_32_out_L&#009;400_32_out_R"/>
+            </assert_stdout>
             <output name="output_bam" ftype="bam">
                 <assert_contents>
-                    <has_size value="3185672" delta="1000"/>
+                    <has_size value="3130923" delta="1000"/>
                 </assert_contents>
             </output>
         </test>

diff --git a/tools/ivar/ivar_trim.xml b/tools/ivar/ivar_trim.xml
@@ -1,4 +1,4 @@
-<tool id="ivar_trim" name="ivar trim" version="@TOOL_VERSION@+galaxy6" profile="@PROFILE@">
+<tool id="ivar_trim" name="ivar trim" version="@TOOL_VERSION@+galaxy0" profile="@PROFILE@">
     <description>Trim reads in aligned BAM</description>
     <macros>
         <import>macros.xml</import>

diff --git a/tools/ivar/ivar_variants.xml b/tools/ivar/ivar_variants.xml
@@ -1,4 +1,4 @@
-<tool id="ivar_variants" name="ivar variants" version="@TOOL_VERSION@+galaxy3" profile="@PROFILE@">
+<tool id="ivar_variants" name="ivar variants" version="@TOOL_VERSION@+galaxy0" profile="@PROFILE@">
     <description>Call variants from aligned BAM file</description>
     <macros>
         <import>macros.xml</import>
@@ -27,7 +27,7 @@
     <inputs>
         <param name="input_bam" type="data" format="bam" label="Bam file" help="Aligned reads, to trim primers and quality"/>
         <param name="ref" type="data" format="fasta" label="Reference"/>
-        <param name="min_qual" argument="-q" type="integer" min="1" value="20" label="Minimum quality score threshold to count base"/>
+        <param name="min_qual" argument="-q" type="integer" min="0" max="255" value="20" label="Minimum quality score threshold to count base"/>
         <param name="min_freq" argument="-t" type="float" min="0" max="1" value="0.03" label="Minimum frequency threshold"/>
         <conditional name="output_format">
             <param name="choice" type="select" label="Output format">
@@ -51,7 +51,7 @@
         <data name="output_variants_tabular" from_work_dir="./variants.tsv" format="tabular" label="${tool.name} tabular output on ${on_string}">
             <filter>output_format['choice'] == 'tabular' or output_format['choice'] == 'tabular_and_vcf'</filter>
             <actions>
-                <action name="column_names" type="metadata" default="REGION,POS,REF,ALT,REF_DP,REF_RV,REF_QUAL,ALT_DP,ALT_RV,ALT_QUAL,ALT_FREQ,TOTAL_DP,PVAL,PASS,GFF_FEATURE,REF_CODON,REF_AA,ALT_CODON,ALT_AA"/>
+                <action name="column_names" type="metadata" default="REGION,POS,REF,ALT,REF_DP,REF_RV,REF_QUAL,ALT_DP,ALT_RV,ALT_QUAL,ALT_FREQ,TOTAL_DP,PVAL,PASS,GFF_FEATURE,REF_CODON,REF_AA,ALT_CODON,ALT_AA,POS_AA"/>
             </actions>
         </data>
         <data name="output_variants_vcf" from_work_dir="./variants.vcf" format="vcf" label="${tool.name} VCF on ${on_string}">

diff --git a/tools/ivar/macros.xml b/tools/ivar/macros.xml
@@ -1,12 +1,12 @@
 <?xml version="1.0"?>
 <macros>
-  <token name="@TOOL_VERSION@">1.3.1</token>
+  <token name="@TOOL_VERSION@">1.3.2</token>
   <token name="@PROFILE@">21.01</token>
   <xml name="requirements">
   <requirements>
       <requirement type="package" version="@TOOL_VERSION@">ivar</requirement>
       <requirement type="package" version="3.10.8">python</requirement>
-      <requirement type="package" version="1.16">samtools</requirement>
+      <requirement type="package" version="1.16.1">samtools</requirement>
       <yield/>
   </requirements>
   </xml>

diff --git a/tools/ivar/test-data/zika/Z52_a.tsv b/tools/ivar/test-data/zika/Z52_a.tsv
@@ -1,17 +1,17 @@
-REGION	POS	REF	ALT	REF_DP	REF_RV	REF_QUAL	ALT_DP	ALT_RV	ALT_QUAL	ALT_FREQ	TOTAL_DP	PVAL	PASS	GFF_FEATURE	REF_CODON	REF_AA	ALT_CODON	ALT_AA
-PRV	350	A	T	101	101	36	5	5	35	0.0471698	106	0.0304989	FALSE	NA	NA	NA	NA	NA
-PRV	722	C	+A	280	234	36	9	0	20	0.0319149	282	0.0349042	FALSE	NA	NA	NA	NA	NA
-PRV	1682	C	T	1097	984	37	34	33	37	0.0300088	1133	7.79053e-10	TRUE	NA	NA	NA	NA	NA
-PRV	1965	T	G	302	113	37	63	25	37	0.172603	365	4.41891e-21	TRUE	NA	NA	NA	NA	NA
-PRV	2702	A	G	31	31	36	1	1	23	0.03125	32	0.507937	FALSE	NA	NA	NA	NA	NA
-PRV	2781	T	G	354	70	37	48	8	36	0.117647	408	4.33605e-16	TRUE	NA	NA	NA	NA	NA
-PRV	2922	C	T	264	0	36	11	0	36	0.04	275	0.000440953	TRUE	NA	NA	NA	NA	NA
-PRV	3148	Y	T	0	0	0	1324	264	36	0.77563	1707	0	TRUE	NA	NA	NA	NA	NA
-PRV	3148	Y	C	0	0	0	381	75	36	0.223199	1707	0	TRUE	NA	NA	NA	NA	NA
-PRV	3295	A	G	1002	1002	35	38	38	33	0.0365385	1040	4.30837e-11	TRUE	NA	NA	NA	NA	NA
-PRV	5680	C	T	27	10	44	8	3	46	0.228571	35	0.000596701	TRUE	NA	NA	NA	NA	NA
-PRV	5723	T	G	31	31	35	1	1	21	0.03125	32	0.507937	FALSE	NA	NA	NA	NA	NA
-PRV	6201	A	G	10	0	35	2	0	38	0.166667	12	0.26087	FALSE	NA	NA	NA	NA	NA
-PRV	6211	T	C	8	0	36	1	0	35	0.111111	9	0.5	FALSE	NA	NA	NA	NA	NA
-PRV	7916	C	T	351	289	36	81	78	37	0.1875	432	5.39655e-27	TRUE	NA	NA	NA	NA	NA
-PRV	9713	C	T	374	0	37	13	0	35	0.0335917	387	0.00010474	TRUE	NA	NA	NA	NA	NA
+REGION	POS	REF	ALT	REF_DP	REF_RV	REF_QUAL	ALT_DP	ALT_RV	ALT_QUAL	ALT_FREQ	TOTAL_DP	PVAL	PASS	GFF_FEATURE	REF_CODON	REF_AA	ALT_CODON	ALT_AA	POS_AA
+PRV	350	A	T	101	101	36	5	5	35	0.0471698	106	0.0304989	FALSE	NA	NA	NA	NA	NA	NA
+PRV	722	C	+A	280	234	36	9	0	20	0.0319149	282	0.0349042	FALSE	NA	NA	NA	NA	NA	NA
+PRV	1682	C	T	1097	984	37	34	33	37	0.0300088	1133	7.79053e-10	TRUE	NA	NA	NA	NA	NA	NA
+PRV	1965	T	G	302	113	37	63	25	37	0.172603	365	4.41891e-21	TRUE	NA	NA	NA	NA	NA	NA
+PRV	2702	A	G	31	31	36	1	1	23	0.03125	32	0.507937	FALSE	NA	NA	NA	NA	NA	NA
+PRV	2781	T	G	354	70	37	48	8	36	0.117647	408	4.33605e-16	TRUE	NA	NA	NA	NA	NA	NA
+PRV	2922	C	T	264	0	36	11	0	36	0.04	275	0.000440953	TRUE	NA	NA	NA	NA	NA	NA
+PRV	3148	Y	T	0	0	0	1324	264	36	0.77563	1707	0	TRUE	NA	NA	NA	NA	NA	NA
+PRV	3148	Y	C	0	0	0	381	75	36	0.223199	1707	0	TRUE	NA	NA	NA	NA	NA	NA
+PRV	3295	A	G	1002	1002	35	38	38	33	0.0365385	1040	4.30837e-11	TRUE	NA	NA	NA	NA	NA	NA
+PRV	5680	C	T	27	10	44	8	3	46	0.228571	35	0.000596701	TRUE	NA	NA	NA	NA	NA	NA
+PRV	5723	T	G	31	31	35	1	1	21	0.03125	32	0.507937	FALSE	NA	NA	NA	NA	NA	NA
+PRV	6201	A	G	10	0	35	2	0	38	0.166667	12	0.26087	FALSE	NA	NA	NA	NA	NA	NA
+PRV	6211	T	C	8	0	36	1	0	35	0.111111	9	0.5	FALSE	NA	NA	NA	NA	NA	NA
+PRV	7916	C	T	351	289	36	81	78	37	0.1875	432	5.39655e-27	TRUE	NA	NA	NA	NA	NA	NA
+PRV	9713	C	T	374	0	37	13	0	35	0.0335917	387	0.00010474	TRUE	NA	NA	NA	NA	NA	NA
diff --git a/tools/ivar/test-data/zika/Z52_a_annotated.tsv b/tools/ivar/test-data/zika/Z52_a_annotated.tsv
@@ -1,17 +1,17 @@
-REGION	POS	REF	ALT	REF_DP	REF_RV	REF_QUAL	ALT_DP	ALT_RV	ALT_QUAL	ALT_FREQ	TOTAL_DP	PVAL	PASS	GFF_FEATURE	REF_CODON	REF_AA	ALT_CODON	ALT_AA
-PRV	350	A	T	101	101	36	5	5	35	0.0471698	106	0.0304989	FALSE	cds-YP_002790881.1	AAA	K	TAA	*
-PRV	722	C	+A	280	234	36	9	0	20	0.0319149	282	0.0349042	FALSE	NA	NA	NA	NA	NA
-PRV	1682	C	T	1097	984	37	34	33	37	0.0300088	1133	7.79053e-10	TRUE	cds-YP_002790881.1	CTG	L	TTG	L
-PRV	1965	T	G	302	113	37	63	25	37	0.172603	365	4.41891e-21	TRUE	cds-YP_002790881.1	GTT	V	GGT	G
-PRV	2702	A	G	31	31	36	1	1	23	0.03125	32	0.507937	FALSE	cds-YP_002790881.1	AGA	R	GGA	G
-PRV	2781	T	G	354	70	37	48	8	36	0.117647	408	4.33605e-16	TRUE	cds-YP_002790881.1	GTG	V	GGG	G
-PRV	2922	C	T	264	0	36	11	0	36	0.04	275	0.000440953	TRUE	cds-YP_002790881.1	ACT	T	ATT	I
-PRV	3148	Y	T	0	0	0	1324	264	36	0.77563	1707	0	TRUE	cds-YP_002790881.1	GAY	X	GAT	D
-PRV	3148	Y	C	0	0	0	381	75	36	0.223199	1707	0	TRUE	cds-YP_002790881.1	GAY	X	GAC	D
-PRV	3295	A	G	1002	1002	35	38	38	33	0.0365385	1040	4.30837e-11	TRUE	cds-YP_002790881.1	GCA	A	GCG	A
-PRV	5680	C	T	27	10	44	8	3	46	0.228571	35	0.000596701	TRUE	cds-YP_002790881.1	TTC	F	TTT	F
-PRV	5723	T	G	31	31	35	1	1	21	0.03125	32	0.507937	FALSE	cds-YP_002790881.1	TGA	*	GGA	G
-PRV	6201	A	G	10	0	35	2	0	38	0.166667	12	0.26087	FALSE	cds-YP_002790881.1	GAA	E	GGA	G
-PRV	6211	T	C	8	0	36	1	0	35	0.111111	9	0.5	FALSE	cds-YP_002790881.1	TGT	C	TGC	C
-PRV	7916	C	T	351	289	36	81	78	37	0.1875	432	5.39655e-27	TRUE	cds-YP_002790881.1	CAG	Q	TAG	*
-PRV	9713	C	T	374	0	37	13	0	35	0.0335917	387	0.00010474	TRUE	cds-YP_002790881.1	CTT	L	TTT	F
+REGION	POS	REF	ALT	REF_DP	REF_RV	REF_QUAL	ALT_DP	ALT_RV	ALT_QUAL	ALT_FREQ	TOTAL_DP	PVAL	PASS	GFF_FEATURE	REF_CODON	REF_AA	ALT_CODON	ALT_AA	POS_AA
+PRV	350	A	T	101	101	36	5	5	35	0.0471698	106	0.0304989	FALSE	POLY:cds-YP_002790881.1	AAA	K	TAA	*	82
+PRV	722	C	+A	280	234	36	9	0	20	0.0319149	282	0.0349042	FALSE	NA	NA	NA	NA	NA	NA
+PRV	1682	C	T	1097	984	37	34	33	37	0.0300088	1133	7.79053e-10	TRUE	POLY:cds-YP_002790881.1	CTG	L	TTG	L	526
+PRV	1965	T	G	302	113	37	63	25	37	0.172603	365	4.41891e-21	TRUE	POLY:cds-YP_002790881.1	GTT	V	GGT	G	620
+PRV	2702	A	G	31	31	36	1	1	23	0.03125	32	0.507937	FALSE	POLY:cds-YP_002790881.1	AGA	R	GGA	G	866
+PRV	2781	T	G	354	70	37	48	8	36	0.117647	408	4.33605e-16	TRUE	POLY:cds-YP_002790881.1	GTG	V	GGG	G	892
+PRV	2922	C	T	264	0	36	11	0	36	0.04	275	0.000440953	TRUE	POLY:cds-YP_002790881.1	ACT	T	ATT	I	939
+PRV	3148	Y	T	0	0	0	1324	264	36	0.77563	1707	0	TRUE	POLY:cds-YP_002790881.1	GAY	X	GAT	D	1014
+PRV	3148	Y	C	0	0	0	381	75	36	0.223199	1707	0	TRUE	POLY:cds-YP_002790881.1	GAY	X	GAC	D	1014
+PRV	3295	A	G	1002	1002	35	38	38	33	0.0365385	1040	4.30837e-11	TRUE	POLY:cds-YP_002790881.1	GCA	A	GCG	A	1063
+PRV	5680	C	T	27	10	44	8	3	46	0.228571	35	0.000596701	TRUE	POLY:cds-YP_002790881.1	TTC	F	TTT	F	1858
+PRV	5723	T	G	31	31	35	1	1	21	0.03125	32	0.507937	FALSE	POLY:cds-YP_002790881.1	TGA	*	GGA	G	1873
+PRV	6201	A	G	10	0	35	2	0	38	0.166667	12	0.26087	FALSE	POLY:cds-YP_002790881.1	GAA	E	GGA	G	2032
+PRV	6211	T	C	8	0	36	1	0	35	0.111111	9	0.5	FALSE	POLY:cds-YP_002790881.1	TGT	C	TGC	C	2035
+PRV	7916	C	T	351	289	36	81	78	37	0.1875	432	5.39655e-27	TRUE	POLY:cds-YP_002790881.1	CAG	Q	TAG	*	2604
+PRV	9713	C	T	374	0	37	13	0	35	0.0335917	387	0.00010474	TRUE	POLY:cds-YP_002790881.1	CTT	L	TTT	F	3203