migration of table-merge

.gitignore

@@ -1,2 +1,3 @@
 tool_test_output.html
 tool_test_output.json
+.Rhistory

tools/tablemerge/.shed.yml

@@ -0,0 +1,11 @@
+categories:
+  - Metabolomics
+description: '[W4M][Utils] Merging dataMatrix with a metadata table.'
+homepage_url: http://workflow4metabolomics.org
+remote_repository_url: https://github.com/workflow4metabolomics/tools-metabolomics
+long_description: 'Part of the W4M project: http://workflow4metabolomics.org / The
+  R script merges the data matrix with a selected metadata file (sample metadata or
+  variable metadata) to obtain a single file.'
+name: tablemerge
+owner: melpetera
+type: unrestricted

tools/tablemerge/README.md

@@ -0,0 +1,51 @@
+# Table Merge
+
+
+Metadata
+-----------
+
+ * **@name**: Table Merge
+ * **@galaxyID**: tablemerge
+ * **@version**: 1.0.2+galaxy1
+ * **@authors**: Original code: Melanie Petera (PFEM - INRAE - MetaboHUB) - Maintainer: Melanie Petera (PFEM - INRAE - MetaboHUB)
+ * **@init date**: 2016, July
+ * **@main usage**: This tool enables the aggregation of a dataMatrix with one of its metadata (sample or variable metadata).
+
+
+Context
+-----------
+
+This tool is provided as one of the [Workflow4Metabolomics](http://workflow4metabolomics.org) Galaxy instance data handling tools. 
+W4M is an international infrastructure providing software tools to process, analyse and annotate metabolomics data. 
+
+User interface is based on the Galaxy platform (homepage: https://galaxyproject.org/). It is an open, web-based platform for data intensive biomedical research. 
+Whether on the free public server or your own instance, you can perform, reproduce, and share complete analyses.
+
+Configuration
+-----------
+
+### Requirement:
+ * R software: version = 4.3.3 recommended
+ * r-w4mrutils = 1.0.0
+
+Technical description
+-----------
+
+Main files:
+
+- tablemerge_script.R: R function (core script)
+- tablemerge.xml: XML wrapper (interface for Galaxy)
+
+
+Services provided
+-----------
+
+ * Help and support: https://community.france-bioinformatique.fr/c/workflow4metabolomics/10
+
+
+
+License
+-----------
+
+ * Cea Cnrs Inria Logiciel Libre License, version 2.1 (CECILL-2.1)
+

tools/tablemerge/tablemerge.xml

@@ -0,0 +1,158 @@
+<tool id="tablemerge" name="Table Merge" version="1.0.2+galaxy1" profile="24.2">
+  <description>Merging dataMatrix with a metadata table</description>
+  <edam_topics>
+    <edam_topic>topic_0769</edam_topic>
+  </edam_topics>
+  <edam_operations>
+    <edam_operation>operation_3436</edam_operation>
+  </edam_operations>
+  <creator>
+    <organization url="https://workflow4metabolomics.org/" email="[email protected]" name="Workflow4Metabolomics" />
+  </creator>
+  <requirements>
+	    <requirement type="package" version="1.0.0">r-w4mrutils</requirement>
+  </requirements>
+
+  <required_files>
+    <include path="tablemerge_script.R" />
+  </required_files>
+
+  <command detect_errors="exit_code"><![CDATA[
+	  Rscript -e 'source("$__tool_directory__/tablemerge_script.R")' -e 'tab.merge("$dataMatrix_in", "$Metadata_in", "$metatype", "$combined_out")'
+  ]]></command>
+
+  <inputs>
+    <param name="dataMatrix_in" type="data" label="Data matrix file" help="" format="tabular" />
+    <param name="Metadata_in" type="data" label="Metadata file" help="Sample metadata or variable metadata" format="tabular" />
+
+    <param name="metatype" label="Type of metadata" type="select" display="radio" help="">
+      <option value="sample">Sample metadata</option>
+      <option value="variable">Variable metadata</option>
+    </param>
+
+  </inputs>
+
+  <outputs>
+    <data name="combined_out" label="Combined_${Metadata_in.name}" format="tabular" ></data>
+  </outputs>
+
+  <tests>
+	  <test>
+      <param name="dataMatrix_in" value="input_TM12_dataMatrix.txt"/>
+      <param name="Metadata_in" value="input_TM1_variableMetadata.txt"/>
+      <param name="metatype" value="variable"/>
+      <output name="combined_out" file="output_TM1_expected.tabular"/>
+	  </test>
+	  <test>
+      <param name="dataMatrix_in" value="input_TM12_dataMatrix.txt"/>
+      <param name="Metadata_in" value="input_TM2_sampleMetadata.txt"/>
+      <param name="metatype" value="sample"/>
+      <output name="combined_out" file="output_TM2_expected.tabular"/>
+	  </test>
+  </tests>
+
+
+  <help><![CDATA[
+
+.. class:: infomark
+
+**Credits** 
+  | **Original tool code & wrapping:** Melanie Petera - PFEM ; INRAE ; MetaboHUB
+  | **Original tool code & wrapping maintainer:** Melanie Petera - PFEM ; INRAE ; MetaboHUB
+
+.. class:: infomark
+
+**Contact:** For any questions or support: community.france-bioinformatique.fr/c/workflow4metabolomics/
+
+---------------------------------------------------
+
+***********
+Table Merge
+***********
+
+===========
+DESCRIPTION
+===========
+
+Merges the data matrix with a selected metadata file (sample metadata or variable metadata) to obtain a single file. 
+
+---------------------------------------------------
+
+==========================
+ALIGNMENT WITH OTHER TOOLS
+==========================
+
+-----------
+INPUT FILES
+-----------
+
++----------------------------+---------+
+| Parameter : num + label    |  Format |
++============================+=========+
+| 1 : Data matrix file       | tabular |
++----------------------------+---------+
+| 2 : Metadata file          | tabular |
++----------------------------+---------+
+
+Data matrix file contains the intensity values of the variables.
+	| 
+
+Metadata file is meant to be chosen from sample metadata and variable metadata files. 
+	| 
+
+.. class:: warningmark
+
+For more information about input files, refer to the corresponding "W4M HowTo" page:
+ | `W4M table format for Galaxy <https://nextcloud.inrae.fr/s/qLkNZRf84QQ5YLY>`_
+ |
+
+------------
+OUTPUT FILES
+------------
+
+Combined_Metadata
+	| tabular output
+	| Corresponds to the input metadata file completed as new columns by the intensities in the input data matrix file
+	|
+
+----------------------------
+EXAMPLE OF WORKFLOW POSITION
+----------------------------
+
+Can be used as a last step before leaving galaxy, to gather informations into a single file. Might be handy to make further analysis with Symca for instance.
+
+---------------------------------------------------
+
+===============
+TOOL PARAMETERS
+===============
+
+Type of metadata
+	| Specify which type of metadata table is given as Metadata file
+	| 
+
+---------------------------------------------------
+
+==========
+CHANGE LOG
+==========
+--------------------------------------
+Version 1.0.2+galaxy1 - September 2022
+--------------------------------------
+INTERNAL MODIFICATION
+- Logic of the tool called from generic repository W4MUtils
+- Update of tool wrapper to comply with good practices
+- Update of help section
+
+-------------------------------
+Version 1.0.2 - September 2022
+-------------------------------
+MINORFIX: removal batch package
+
+
+
+	]]></help>
+	<citations>
+    <citation type="doi">10.1093/bioinformatics/btu813</citation>
+  </citations>
+</tool>

tools/tablemerge/tablemerge_script.R

@@ -0,0 +1,82 @@
+################################################################################################
+# TABLE MERGE                                                                                  #
+#                                                                                              #
+# User: Galaxy                                                                                 #
+# Starting date: 16-04-2015                                                                    #
+# V-0.1: First version of merge code                                                           #
+# V-0.2: Addition of data check and handling of special characters                             #
+#                                                                                              #
+#                                                                                              #
+# Input files: dataMatrix ; Metadata file                                                      #
+# Output files: dataMatrix ; Metadata file                                                     #
+#                                                                                              #
+# Dependencies: RcheckLibrary.R ; miniTools.R                                                  #
+#                                                                                              #
+################################################################################################
+library(W4MRUtils)
+
+# Parameters (for dev)
+if (FALSE) {
+    DM.name <- "dataMatrix_CleanIons_CleanEch.txt"
+    meta.name <- "sampleMetadata_CleanEch.txt"
+    metype <- "sample"
+    output <- "Combined_${Metadata_in.name}"
+}
+
+
+
+tab.merge <- function(DM.name, meta.name, metype, output) {
+    # This function allows to merge the dataMatrix with one metadata table.
+    #
+    # Parameters:
+    # - DM.name, meta.name: dataMatrix and metadata files' access respectively
+    # - metype: "sample" or "variable" depending on metadata content
+    # - output: output file's access
+
+
+    # Input --------------------------------------------------------------
+    DM <- read.table(DM.name, header = TRUE, sep = "\t", check.names = FALSE)
+    meta <- read.table(meta.name, header = TRUE, sep = "\t", check.names = FALSE, colClasses = "character", comment.char = "")
+
+    # Table match check
+    table.check <- match2(DM, meta, metype)
+    check_err(table.check)
+
+    # StockID
+    meta.id <- stock_id(DM, meta, metype)
+    DM <- meta.id$dataMatrix
+    meta <- meta.id$Metadata
+    meta.id <- meta.id$id.match
+
+
+    # Merging tables -----------------------------------------------------
+
+    if (metype == "sample") {
+        ori.DM <- DM
+        rownames(DM) <- DM[, 1]
+        DM <- DM[, -1]
+        DM <- t(DM)
+        DM <- data.frame(sample = row.names(DM), DM, check.names = FALSE)
+        rownames(DM) <- NULL
+    }
+
+    comb.data <- merge(x = meta, y = DM, by.x = 1, by.y = 1)
+
+
+    # Output -------------------------------------------------------------
+
+    # Getting back original identifiers
+    if (metype == "sample") {
+        id.ori <- reproduce_id(ori.DM, comb.data, metype, meta.id)
+    } else {
+        id.ori <- reproduce_id(DM, comb.data, metype, meta.id)
+    }
+    comb.data <- id.ori$Metadata
+
+    # Writing the table
+    write.table(comb.data, output, sep = "\t", quote = FALSE, row.names = FALSE)
+} # End of tab.merge
+
+
+# Typical function call
+# tab.merge(DM.name,meta.name,metype,output)

tools/tablemerge/test-data/input_TM12_dataMatrix.txt

@@ -0,0 +1,18 @@
+data	j 785	y54j 68y4j6	5-6 4	hrh	5h -	3	t564	t54 66	y6y	t6 5h	(5y	g51	(	6	98 j7-0	06654h	60
+5d_-kkcùf	0.356426723610756	0.380152310071702	0.0306944207412024	0.334137638848017	0.298147485608469	0.975073793297568	0.129099504745855	0.973451663994064	0.334792719284834	0.0751565260418877	0.322897933010729	0.129033714279026	0.722966330196726	0.580654367692078	0.644502528016206	0.352827235332827	0.511778286438001
+npèt	0.801750949101246	0.535593104115636	0.258351312473067	0.599475573145688	0.0763319665667417	0.504558103160623	0.367963830290116	0.440540211857668	0.158066366765388	0.609574253877002	0.550590276177951	0.232864384033621	0.637065775537391	0.424098276140436	0.662053737304139	0.0217541227637467	0.299053946391647
+5PY4(*3	0.875199970420953	0.0825428101366147	0.253659010703906	0.507762246807195	0.856444177031262	0.291594962324086	0.616538655402613	0.388475672176377	0.914254939740854	0.449607330807756	0.628469388494906	0.436558212905713	0.951198610265672	0.956459387015831	0.0113483457928278	0.611559621193868	0.127393349941781
+k	0.235214515530521	0.50917637648828	0.0303152651414629	0.305595242012361	0.470145996588881	0.356776179776951	0.569840649303474	0.720070419680388	0.875026990189028	0.97246313990283	0.936423978578332	0.810603455355391	0.903046784946495	0.667091730522105	0.0282601148338295	0.276666638740528	0.747503427028951
+t v^ptok!R/;	0.251801918471682	0.871383805557867	0.285183681272061	0.552739436194454	0.863766301215097	0.182656918760194	0.977603710051043	0.204845724382198	0.751690583305025	0.751475804910098	0.938399878874186	0.218163132174348	0.523463794277572	0.987156468237196	0.630938391969655	0.495061607430081	0.815474132321848
+ojt*	0.907412839750932	0.857498273276021	0.370137361785406	0.76678484779183	0.213299689296413	0.521214352529346	0.0851731095099456	0.868227767553325	0.0711706993343434	0.70186429888193	0.700568035764899	0.7467824290166	0.570332959002042	0.23270290523468	0.126423339798572	0.816262268500797	0.466363859676536
+lmtki	0.344288940919619	0.308219131305663	0.218449898667526	0.700869668247368	0.932538878977527	0.396198994630489	0.710306820874376	0.124235068279782	0.56608643523601	0.887647287670482	0.0375011938139768	0.282378766400105	0.757492159558571	0.506567258021703	0.990910788272054	0.326205008250378	0.61003856679129
+1h6-(587 t	0.308709890328272	0.717202505490233	0.777934118812417	0.604807545422864	0.588458968776734	0.940112596977594	0.788662314418432	0.133046934771924	0.891405172730758	0.915119678712577	0.401136527896346	0.996068607925829	0.049077648050296	0.157657286859101	0.598183619914656	0.439531332356748	0.11995300831499
+r154 (54 86	0.0861033910635205	0.243889357950567	0.848801588667359	0.972183114768299	0.311829870889657	0.434854970574638	0.0240613698880671	0.273335646139236	0.902686361056168	0.73479376111027	0.232132726965829	0.471097756642369	0.511949019774828	0.934158039531663	0.60125450357768	0.972297241140637	0.63149263296722
+(h4	0.336373953934809	0.521412085356266	0.281632887739949	0.67004516763179	0.0199332367594847	0.867732114278491	0.973477443505332	0.883197938525362	0.334509105447652	0.793947578560618	0.465167375556954	0.508757674191867	0.983758014396725	0.100582744588671	0.501301982240787	0.789637338939369	0.967382850752139
+2	0.65083934802851	0.915695024588	0.0376667088550816	0.514003854683931	0.274761574286474	0.652108809243866	0.689393057486628	0.262205073769319	0.871958663956254	0.709490850507154	0.854477673591409	0.672918329289054	0.141693817710263	0.00543057257281332	0.273084303937691	0.39570748778882	0.303196826092465
+m(yk	0.798331012769621	0.806128809071498	0.519974578571807	0.0237856823099186	0.396426673475168	0.75979784245693	0.0589336105092753	0.763144123705766	0.16086904822931	0.301000258395631	0.321186481311644	0.0091434029302796	0.0346149352313591	0.276366137673702	0.686099471259438	0.0548240244354505	0.909361338008149
+-546 -3 -3	0.981561422839383	0.755973696318293	0.328878596436459	0.233341270249666	0.14569201922932	0.458105216271765	0.191023237667199	0.775603857680909	0.146769440956804	0.86812372109824	0.601678275863019	0.782019307745586	0.955614654474782	0.836058192741214	0.454550085076576	0.181788519803465	0.325277631599655
+nz^à  t	0.920089977342539	0.249031996586174	0.105577675363886	0.805216649434891	0.505774902707566	0.619612431793644	0.337662179244335	0.672829320507873	0.387228581340173	0.209795853201976	0.0403285073621635	0.540614252108851	0.43362855410287	0.160204078494507	0.924783730297701	0.116837628312753	0.58763473928668
+kzjù	0.418651623550921	0.668434005412235	0.845525871017258	0.567468950919516	0.111441846749533	0.411424602517146	0.0743349602270699	0.583730221710347	0.0105008018078939	0.918789674284249	0.808643595994088	0.33578972219198	0.536848200622412	0.898842910125025	0.333954421105444	0.0390509357521919	0.159857705429294
+. Glk(ôepjg$	0.3266193089274	0.806668730531573	0.87265879590923	0.934940869880726	0.548967615908753	0.250539517744596	0.75436043048449	0.597444122305604	0.343955002787363	0.560987092777944	0.134929631392942	0.958047956861493	0.870429681664852	0.66286672214456	0.441523248613101	0.437073092741238	0.265031623006715
+rmlgj*lr ,	0.462672931523031	0.74679446664976	0.151929368261195	0.165870989887449	0.90931621240424	0.423706631494886	0.337742032589001	0.184993498717115	0.0792523932129463	0.071353352835237	0.525247607449536	0.647744088759121	0.158725526449032	0.203927461340644	0.20055683200405	0.897053444943847	0.506954538974894

tools/tablemerge/test-data/input_TM1_variableMetadata.txt

@@ -0,0 +1,18 @@
+variable	A	B	C	D
+5d_-kkcùf	0	a	ze$ùj	0
+npèt	1	b	 g	0
+5PY4(*3	2	c	 15j-è,	0
+k	3	a	6jè	1
+t v^ptok!R/;	4	b	8j4,	1
+ojt*	5	c	5èj1	1
+lmtki	6	a	j84	2
+1h6-(587 t	7	b	j4e54	2
+r154 (54 86	8	c	j	2
+(h4	9	a	54j6@44èu6(è4	3
+2	10	b	(51h	3
+m(yk	11	c	8-1	3
+-546 -3 -3	12	a	hy-81u	4
+nz^à  t	13	b	zs5-hj1	4
+kzjù	14	c	861h	4
+. Glk(ôepjg$	15	a	(8168(h1	9
+rmlgj*lr ,	16	b	8hjè1	9

tools/tablemerge/test-data/input_TM2_sampleMetadata.txt

@@ -0,0 +1,18 @@
+sample	A	B	C	D	E	F
+j 785	1	y	z	1.2	F	jkh52è16
+y54j 68y4j6	3	y	e	1.6	F	1è5hj4-
+5-6 4	5	y	z	1.5	F	-
+hrh	9	y	e	6.3	F	è1484j
+5h -	8	y	z	6.9	F	h3254
+3	7	y	e	0	F	
+t564	4	y	z	2	F	th651
+t54 66	5	y	e	6	F	h4+
+y6y	6	y	z	8	F	he+65+(
+t6 5h	8	y	e	4	F	h56h4+
+(5y	5	y	z	5	F	
+g51	1	y	e	7	F	
+(	2	y	z	6	F	5h4te+4
+6	6	y	e	5	F	yhe6
+98 j7-0	4	y	z	3	F	theh
+06654h	4	y	e	7	F	56
+60	2	y	z	3	F