Migration: isa-extractor

tools/isa-extractor/.gitignore

@@ -0,0 +1,4 @@
+planemo-venv
+tool_test_output.*
+dist
+.Rproj.user

tools/isa-extractor/.travis.yml

@@ -0,0 +1,11 @@
+dist: bionic
+
+notifications:
+ email:
+  recipients:
+   - [email protected]
+
+script:
+    - make test
+    - make plint
+    - make ptest

tools/isa-extractor/Makefile

@@ -0,0 +1,46 @@
+SCRIPT_NAME=extract-from-isa
+REPOS_NAME=isaextractor
+TOOLS_XML=$(wildcard isa2*.xml)
+
+all:
+
+test:
+	make -C $@
+
+planemo-venv/bin/planemo: planemo-venv
+	. planemo-venv/bin/activate && pip install --upgrade pip setuptools
+	. planemo-venv/bin/activate && pip install planemo
+
+planemo-venv:
+	virtualenv -p python2.7 $@
+
+plint: planemo-venv/bin/planemo
+	. planemo-venv/bin/activate && planemo lint $(TOOLS_XML)
+
+ptest: planemo-venv/bin/planemo
+	. planemo-venv/bin/activate && planemo test --conda_dependency_resolution --galaxy_branch release_19.01 $(TOOLS_XML)
+
+dist/$(REPOS_NAME)/:
+	mkdir -p $@
+	cp -Lr README.md $(SCRIPT_NAME) $(TOOLS_XML) test-data $@
+
+ptesttoolshed_diff: dist/$(REPOS_NAME)/ planemo-venv/bin/planemo
+	. planemo-venv/bin/activate && cd $< && planemo shed_diff --shed_target testtoolshed
+
+ptesttoolshed_update: dist/$(REPOS_NAME)/ planemo-venv/bin/planemo
+	. planemo-venv/bin/activate && cd $< && planemo shed_update --check_diff --shed_target testtoolshed
+
+ptoolshed_diff: dist/$(REPOS_NAME)/ planemo-venv/bin/planemo
+	. planemo-venv/bin/activate && cd $< && planemo shed_diff --shed_target toolshed
+
+ptoolshed_update: dist/$(REPOS_NAME)/ planemo-venv/bin/planemo
+	. planemo-venv/bin/activate && cd $< && planemo shed_update --check_diff --shed_target toolshed
+
+clean:
+	$(RM) -r $(HOME)/.planemo
+	$(RM) -r planemo-venv
+	make -C test $@
+	$(RM) tool_test_output.*
+	$(RM) -r dist
+
+.PHONY:	all clean test planemolint planemotest

tools/isa-extractor/README.md

@@ -0,0 +1,12 @@
+# ISA extractor
+
+[![Build Status](https://travis-ci.org/workflow4metabolomics/isa-extractor.svg?branch=master)](https://travis-ci.org/workflow4metabolomics/isa-extractor)
+
+Extract collections of raw data files (mzML, mzXML, netCDF, mzData or nmrML) from an ISA study.
+
+## Updates
+
+### 1.3.0
+
+ * Wrote planemo tests for all tools.
+ * Run tests in Travis-CI.

tools/isa-extractor/extract-from-isa

@@ -0,0 +1,130 @@
+#!/bin/bash
+# vi: fdm=marker
+
+# Constants {{{1
+################################################################
+
+PROG_NAME=$(basename $0)
+PROG_PATH=$(dirname $0)
+YES=yes
+NO=no
+
+# Global variables {{{1
+################################################################
+
+DEBUG=0
+EXT=
+INPUT_DIR=
+OUTPUT_DIR=
+SYMLINK=$NO
+
+# Print help {{{1
+################################################################
+
+function print_help {
+	echo "Usage: $PROG_NAME -i <ISA_DIR> -e <EXT> -o <OUTPUT_DIR>"
+	echo
+	echo "Extract files with a given extension from ISA-Tab archives into a collection."
+	echo
+	echo "Options:"
+	echo "   -e, --ext             EXT   The extension of the files to find."
+	echo "   -h, --help                  Print this help message."
+	echo "   -i, --input           DIR   Input directory containing ISA archive."
+	echo "   -o, --output          DIR   Set the output directory to use."
+	echo "   -s, --symlink               Create symbolic links instead of copying files."
+}
+
+# Error {{{1
+################################################################
+
+function error {
+
+	local msg=$1
+
+	echo "ERROR: $msg" >&2
+
+	exit 1
+}
+
+# Print debug msg {{{1
+################################################################
+
+function print_debug_msg {
+
+	local dbglvl=$1
+	local dbgmsg=$2
+
+	[ $DEBUG -ge $dbglvl ] && echo "[DEBUG] $dbgmsg" >&2
+}
+
+# Read args {{{1
+################################################################
+
+function read_args {
+
+	local args="$*" # save arguments for debugging purpose
+
+	# Read options
+	while true ; do
+		shift_count=1
+		case $1 in
+			-e|--ext)               EXT="$2" ; shift_count=2 ;;
+			-g|--debug)             DEBUG=$((DEBUG + 1)) ;;
+			-h|--help)              print_help ; exit 0 ;;
+			-i|--input)             INPUT_DIR="$2" ; shift_count=2 ;;
+			-o|--output)            OUTPUT_DIR="$2" ; shift_count=2 ;;
+			-s|--symlink)           SYMLINK=$YES ;;
+			-) error "Illegal option $1." ;;
+			--) error "Illegal option $1." ;;
+			--*) error "Illegal option $1." ;;
+			-?) error "Unknown option $1." ;;
+			-[^-]*) split_opt=$(echo $1 | sed 's/^-//' | sed 's/\([a-zA-Z]\)/ -\1/g') ; set -- $1$split_opt "${@:2}" ;;
+			*) break
+		esac
+		shift $shift_count
+	done
+	shift $((OPTIND - 1))
+
+	# Debug
+	print_debug_msg 1 "Arguments are : $args"
+
+	# Check input params
+	[[ $# -eq 0 ]] || error "No remaining arguments are allowed."
+	[[ -n $INPUT_DIR ]] || error "You must specify an input directory, using -i option."
+	[[ -d $INPUT_DIR ]] || error "\"$INPUT_DIR\" is not a valid directory."
+	[[ -n $OUTPUT_DIR ]] || error "You must specify an output directory, using -o option."
+	[[ ! -e $OUTPUT_DIR ]] || error "\"$OUTPUT_DIR\" already exists."
+	[[ -n $EXT ]] || error "You must specify the extension of the files you are looking for, with the -e option."
+}
+
+# MAIN {{{1
+################################################################
+
+read_args "$@"
+
+# Create output directory
+print_debug_msg 1 "Create output directory \"$OUTPUT_DIR\"."
+mkdir -p "$OUTPUT_DIR"
+
+# Find files to extract
+print_debug_msg 1 "Find \"$EXT\" files to extract in \"$INPUT_DIR\"."
+files_to_extract=$(mktemp -t tmp.XXXXXX)
+find "$(realpath $INPUT_DIR)" -iname "*.$EXT" >files_to_extract
+print_debug_msg 1 "Files to extract:"
+if [[ $DEBUG -ge 1 ]] ; then
+	cat files_to_extract >&2
+fi
+
+# Extract files
+if [[ $SYMLINK == $YES ]] ; then
+	print_debug_msg 1 "Create symbolic links of all \"$EXT\" files to extract into \"$OUTPUT_DIR\"."
+	xargs -I % ln -s % "$OUTPUT_DIR" <files_to_extract
+else
+	print_debug_msg 1 "Copy all \"$EXT\" files to extract to \"$OUTPUT_DIR\"."
+	xargs -I % cp % "$OUTPUT_DIR" <files_to_extract
+fi
+rm files_to_extract
+print_debug_msg 1 "Files extracted:"
+if [[ $DEBUG -ge 1 ]] ; then
+	ls -1 "$OUTPUT_DIR" >&2
+fi

tools/isa-extractor/isa-extractor.Rproj

@@ -0,0 +1,18 @@
+Version: 1.0
+
+RestoreWorkspace: Default
+SaveWorkspace: Default
+AlwaysSaveHistory: Default
+
+EnableCodeIndexing: Yes
+UseSpacesForTab: Yes
+NumSpacesForTab: 2
+Encoding: UTF-8
+
+RnwWeave: Sweave
+LaTeX: pdfLaTeX
+
+AutoAppendNewline: Yes
+StripTrailingWhitespace: Yes
+
+BuildType: Makefile

tools/isa-extractor/isa2mzdata.xml

@@ -0,0 +1,77 @@
+<!-- vi: se fdm=marker : -->
+<tool id="isa2mzdata" name="ISA to mzData" version="1.3.0">
+
+	<description>Extract mzData files from an ISA dataset and output a collection of mzData dataset.</description>
+
+	<!-- Command {{{1 -->
+
+	<command><![CDATA[
+		## @@@BEGIN_CHEETAH@@@
+
+		$__tool_directory__/extract-from-isa
+		-i "$isa.extra_files_path"
+		-e mzData
+		-o mzData
+
+		## @@@END_CHEETAH@@@
+	]]></command>
+
+	<!-- Inputs {{{1 -->
+
+	<inputs>
+		<param name="isa" label="ISA" type="data" format="isa-tab"/>
+	</inputs>
+
+	<!-- Outputs {{{1 -->
+
+    <outputs>
+		<collection name="mzData" type="list" label="mzData files">
+			<discover_datasets pattern="(?P&lt;designation&gt;.+)\.[mM][zZ][dD][aA][tT][aA]$" directory="mzData" format="mzdata"/>
+		</collection>
+	</outputs>
+
+	<!-- Tests {{{1 -->
+	<tests>
+		<test>
+			<param name="isa" value="mzdata_study.zip" ftype="isa-tab"/>
+			<output_collection name="mzData" type="list" count="1">
+				<element name="empty" file="mzdata_study_output/empty.mzData" ftype="mzdata"/>
+			</output_collection>
+		</test>
+	</tests>
+
+	<!-- Help {{{1 -->
+    <help>
+<!-- @@@BEGIN_RST@@@ -->
+
+====================
+ISA to mzData
+====================
+
+Extract mzData files contained inside an ISA archive.
+
+-----
+Input
+-----
+
+ISA dataset
+===========
+
+The ISA-Tab dataset from which to extract the files.
+
+------
+Output
+------
+
+The output is a collection of mzData files.
+
+<!-- @@@END_RST@@@ -->
+    </help>
+
+	<!-- Citations {{{1 -->
+    <citations>
+        <citation type="doi">10.1038/ng.1054</citation> <!-- ISA -->
+        <citation type="doi">10.1093/bioinformatics/btu813</citation> <!-- W4M -->
+    </citations>
+
+</tool>

tools/isa-extractor/isa2mzml.xml

@@ -0,0 +1,77 @@
+<!-- vi: se fdm=marker : -->
+<tool id="isa2mzml" name="ISA to mzML" version="1.3.0">
+
+	<description>Extract mzML files from an ISA dataset and output a collection of mzML dataset.</description>
+
+	<!-- Command {{{1 -->
+
+	<command><![CDATA[
+		## @@@BEGIN_CHEETAH@@@
+
+		$__tool_directory__/extract-from-isa
+		-i "$isa.extra_files_path"
+		-e mzML
+		-o mzML
+
+		## @@@END_CHEETAH@@@
+	]]></command>
+
+	<!-- Inputs {{{1 -->
+
+	<inputs>
+		<param name="isa" label="ISA" type="data" format="isa-tab"/>
+	</inputs>
+
+	<!-- Outputs {{{1 -->
+
+    <outputs>
+		<collection name="mzML" type="list" label="mzML files">
+			<discover_datasets pattern="(?P&lt;designation&gt;.+)\.[mM][zZ][mM][lL]$" directory="mzML" format="mzml"/>
+		</collection>
+	</outputs>
+
+	<!-- Tests {{{1 -->
+	<tests>
+		<test>
+			<param name="isa" value="mzml_study.zip" ftype="isa-tab"/>
+			<output_collection name="mzML" type="list" count="1">
+				<element name="empty" file="mzml_study_output/empty.mzML" ftype="mzml"/>
+			</output_collection>
+		</test>
+	</tests>
+
+	<!-- Help {{{1 -->
+    <help>
+<!-- @@@BEGIN_RST@@@ -->
+
+====================
+ISA to mzML
+====================
+
+Extract mzML files contained inside an ISA archive.
+
+-----
+Input
+-----
+
+ISA dataset
+===========
+
+The ISA-Tab dataset from which to extract the files.
+
+------
+Output
+------
+
+The output is a collection of mzML files.
+
+<!-- @@@END_RST@@@ -->
+    </help>
+
+	<!-- Citations {{{1 -->
+    <citations>
+        <citation type="doi">10.1038/ng.1054</citation> <!-- ISA -->
+        <citation type="doi">10.1093/bioinformatics/btu813</citation> <!-- W4M -->
+    </citations>
+
+</tool>