From 8640076f561b1674648ebfbab557764b8dfcbb76 Mon Sep 17 00:00:00 2001
From: hugolefeuvre <hugo.lefeuvre@univ-nantes.fr>
Date: Thu, 14 Nov 2024 15:52:01 +0100
Subject: [PATCH] Correction for the extension (only .fastq or .fastq.gz,
 regardless of the original extension) and no more output corresponding to the
 log

---
 tools/groot/groot.xml           | 22 ++--------
 tools/groot/test-data/groot.log | 77 ---------------------------------
 2 files changed, 4 insertions(+), 95 deletions(-)
 delete mode 100644 tools/groot/test-data/groot.log

diff --git a/tools/groot/groot.xml b/tools/groot/groot.xml
index 7551e419b05..59ac687e668 100644
--- a/tools/groot/groot.xml
+++ b/tools/groot/groot.xml
@@ -2,7 +2,7 @@
     <description> align reads to references and weight variation graphs and generate report</description>
     <macros>
         <token name="@TOOL_VERSION@">1.1.2</token>
-        <token name="@VERSION_SUFFIX@">0</token>
+        <token name="@VERSION_SUFFIX@">1</token>
         <token name="@PROFILE@">22.05</token>
     </macros>
     <xrefs>
@@ -13,7 +13,7 @@
     </requirements>
     <command detect_errors="exit_code"><![CDATA[
 set -x pipefail;
-#set $extension = str($fastq.ext)
+#set $extension = "fastq.gz" if str($fastq.ext).endswith(".gz") else "fastq"
 ln -s -f '${fastq}' 'input.${extension}' &&
 groot index
     --msaDir '$groot_db_select.fields.path'
@@ -31,7 +31,6 @@ groot align
     --indexDir 'grootIndex'
     --contThresh $contThresh
     --minKmerCov $minKmerCov
-    --log '$log'
     $noAlign
     --processors "\${GALAXY_SLOTS:-1}"
 
@@ -69,11 +68,10 @@ groot align
         </section>
     </inputs>
     <outputs>
-        <data name="log" format="txt" label="${tool.name} on ${on_string}: Log"/>
         <data name="report_out" format="tabular" label="${tool.name} on ${on_string}: Report"/>
     </outputs>
     <tests>
-        <test expect_num_outputs="2">
+        <test expect_num_outputs="1">
             <param name="fastq" value="bla-b7-150bp-5x.fq" ftype="fastq"/>
             <section name="index">
                 <param name="groot_db_select" value="resfinder.90" />
@@ -102,14 +100,8 @@ groot align
                     <has_n_columns n="4"/>
                 </assert_contents>
             </output>
-            <output name="log" ftype="txt">
-                <assert_contents>
-                    <has_text text="version"/>
-                    <has_text text="1.1.2"/>
-                </assert_contents>
-            </output>
         </test>
-        <test expect_num_outputs="2">
+        <test expect_num_outputs="1">
             <param name="fastq" value="argannot_light-150bp-10000-reads.fq.gz" ftype="fastq.gz"/>
             <section name="index">
                 <param name="groot_db_select" value="arg-annot.90" />
@@ -137,12 +129,6 @@ groot align
                     <has_n_columns n="4"/>
                 </assert_contents>
             </output>
-            <output name="log" ftype="txt">
-                <assert_contents>
-                    <has_text text="version"/>
-                    <has_text text="1.1.2"/>
-                </assert_contents>
-            </output>
         </test>
     </tests>
     <help><![CDATA[
diff --git a/tools/groot/test-data/groot.log b/tools/groot/test-data/groot.log
deleted file mode 100644
index a9012c05ae5..00000000000
--- a/tools/groot/test-data/groot.log
+++ /dev/null
@@ -1,77 +0,0 @@
-2024/10/15 11:32:06 i am groot (version 1.1.2)
-2024/10/15 11:32:06 starting the index subcommand
-2024/10/15 11:32:06 checking parameters...
-2024/10/15 11:32:06 	directory containing MSA files: arg-annot.90
-2024/10/15 11:32:06 	number of MSA files: 12
-2024/10/15 11:32:06 	processors: 1
-2024/10/15 11:32:06 	k-mer size: 41
-2024/10/15 11:32:06 	sketch size: 21
-2024/10/15 11:32:06 	graph window size: 150
-2024/10/15 11:32:06 	num. partitions: 8
-2024/10/15 11:32:06 	max. K: 4
-2024/10/15 11:32:06 	max. sketch span: 30
-2024/10/15 11:32:06 initialising indexing pipeline...
-2024/10/15 11:32:06 	initialising the processes
-2024/10/15 11:32:06 	connecting data streams
-2024/10/15 11:32:06 	number of processes added to the indexing pipeline: 3
-2024/10/15 11:32:06 creating graphs, sketching traversals and indexing...
-2024/10/15 11:32:07 	number of groot graphs built: 12
-2024/10/15 11:32:07 		graphs sketched: 12
-2024/10/15 11:32:07 		graph windows processed: 35387
-2024/10/15 11:32:07 		mean approximate distinct sketches per graph: 29.20%
-2024/10/15 11:32:07 	number of sketches added to the LSH Ensemble index: 5155
-2024/10/15 11:32:07 writing index files in "grootIndex"...
-2024/10/15 11:32:07 finished in 685.70958ms
-2024/10/15 11:32:39 i am groot (version 1.1.2)
-2024/10/15 11:32:39 starting the sketch subcommand
-2024/10/15 11:32:39 checking parameters...
-2024/10/15 11:32:39 terminated
-
-ERROR --> file does not exist: argannot-150bp-10000-reads.fq.gz
-
-2024/10/15 11:33:04 i am groot (version 1.1.2)
-2024/10/15 11:33:04 starting the sketch subcommand
-2024/10/15 11:33:04 checking parameters...
-2024/10/15 11:33:04 	minimum k-mer coverage: 1
-2024/10/15 11:33:04 	processors: 1
-2024/10/15 11:33:04 	input file: argannot_light-150bp-10000-reads.fq.gz
-2024/10/15 11:33:04 loading the index information...
-2024/10/15 11:33:04 	k-mer size: 41
-2024/10/15 11:33:04 	sketch size: 21
-2024/10/15 11:33:04 	window size used in indexing: 150
-2024/10/15 11:33:04 loading the graphs...
-2024/10/15 11:33:04 	number of variation graphs: 12
-2024/10/15 11:33:04 rebuilding the LSH Ensemble...
-2024/10/15 11:33:04 	containment threshold: 0.99
-2024/10/15 11:33:04 initialising alignment pipeline...
-2024/10/15 11:33:04 	initialising the processes
-2024/10/15 11:33:04 	connecting data streams
-2024/10/15 11:33:04 	number of processes added to the alignment pipeline: 5
-2024/10/15 11:33:04 now streaming reads...
-2024/10/15 11:33:04 	number of reads received from input: 5000
-2024/10/15 11:33:04 	mean read length: 150
-2024/10/15 11:33:04 	number of reads sketched: 5000
-2024/10/15 11:33:04 	total number of unmapped reads: 4873
-2024/10/15 11:33:04 	total number of mapped reads: 127
-2024/10/15 11:33:04 		mapped to one graph: 127
-2024/10/15 11:33:04 		mapped to multiple graphs: 0
-2024/10/15 11:33:04 	total number of exact alignments: 954
-2024/10/15 11:33:04 processing graphs...
-2024/10/15 11:33:04 	total number of k-mers projected onto graphs: 10890
-2024/10/15 11:33:04 	graph 7 has 4 remaining paths after weighting and pruning
-2024/10/15 11:33:04 	- [*argannot~~~(Bla)cfxA2~~~AF504910:1-966]
-2024/10/15 11:33:04 	- [argannot~~~(Bla)cfxA4~~~AY769933:1-966]
-2024/10/15 11:33:04 	- [argannot~~~(Bla)cfxA5~~~AY769934:28-993]
-2024/10/15 11:33:04 	- [argannot~~~(Bla)cfxA~~~U38243:150-1115]
-2024/10/15 11:33:04 	total number of graphs pruned: 12
-2024/10/15 11:33:04 	total number of graphs remaining: 1
-2024/10/15 11:33:04 	total number of possible haplotypes found: 4
-2024/10/15 11:33:04 saving graphs...
-2024/10/15 11:33:04 finished in 223.277576ms
-2024/10/15 11:33:15 i am groot (version 1.1.2)
-2024/10/15 11:33:15 starting the report subcommand
-2024/10/15 11:33:15 checking parameters...
-2024/10/15 11:33:15 	BAM file: groot.bam
-2024/10/15 11:33:15 	coverage cutoff: 0.97
-2024/10/15 11:33:15 	processors: 1
-2024/10/15 11:33:15 finished