[WIP] Additions to avoid hard coded paths in atlas for condensed SDRF calling

applyAllFixesForExperiment.sh

@@ -0,0 +1,80 @@
+#!/usr/bin/env bash
+
+
+usage() { echo "Usage: $0 <accession> path-to-atlas-exps" 1>&2; }
+
+expAcc=$1
+
+if [ -z "${expAcc}" ]; then
+  usage
+  exit 1
+fi
+
+scriptDir=$(cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )
+
+exp=$1
+echo "Applying fixes for $exp ..."
+
+# Applies fixes encoded in $fixesFile to $exp.$fileTypeToBeFixed.txt
+applyFixes() {
+  exp=$1
+  fixesFile=$2
+  fileTypeToBeFixed=$3
+
+  # Apply factor type fixes in ${fileTypeToBeFixed} file
+  for l in $(cat $scriptDir/$fixesFile | sed 's|[[:space:]]*$||g');
+  do
+    if [ ! -s "$exp/$exp.${fileTypeToBeFixed}" ]; then
+      echo "ERROR: $exp/$exp.${fileTypeToBeFixed} not found or is empty" >&2
+      return 1
+    fi
+    echo $l | grep -P '\t' > /dev/null
+    if [ $? -ne 0 ]; then
+      echo  "WARNING: line: '$l' in automatic_fixes_properties.txt is missing a tab character - not applying the fix "
+    fi
+    correct=`echo $l | awk -F"\t" '{print $1}'`
+    toBeReplaced=`echo $l | awk -F"\t" '{print $2}' | sed 's/[^-A-Za-z0-9_ ]/\\\&/g'`
+
+    if [ "$fixesFile" == "automatic_fixes_properties.txt" ]; then
+      # in sdrf or condensed-sdrv fix factor/characteristic types only
+      #if [ "$fileTypeToBeFixed" == "sdrf.txt" ]; then
+      #perl -pi -e "s|\[${toBeReplaced}\]|[${correct}]|g" $exp/$exp.${fileTypeToBeFixed}
+      if [ "$fileTypeToBeFixed" == "condensed-sdrf.tsv" ]; then
+        # In condensed-sdrf, the factor/characteristic type is the penultimate column - so tabs on both sides
+        perl -pi -e "s|\t${toBeReplaced}\t|\t${correct}\t|g" $exp/$exp.${fileTypeToBeFixed}
+      else
+        # idf
+        perl -pi -e "s|\t${toBeReplaced}\t|\t${correct}\t|g" $exp/$exp.${fileTypeToBeFixed}
+        perl -pi -e "s|\t${toBeReplaced}$|\t${correct}|g" $exp/$exp.${fileTypeToBeFixed}
+      fi
+    elif [ "$fixesFile" == "automatic_fixes_values.txt" ]; then
+      #if [ "$fileTypeToBeFixed" == "sdrf.txt" ]; then
+      #perl -pi -e "s|\t${toBeReplaced}\t|\t${correct}\t|g" $exp/$exp.${fileTypeToBeFixed}
+      #perl -pi -e "s|\t${toBeReplaced}$|\t${correct}|g" $exp/$exp.${fileTypeToBeFixed}
+      if [ "$fileTypeToBeFixed" == "condensed-sdrf.tsv" ]; then
+        # In condensed-sdrf, the factor/characteristic value is the last column - so tab on the left and line ending on the right
+        perl -pi -e "s|\t${toBeReplaced}$|\t${correct}|g" $exp/$exp.${fileTypeToBeFixed}
+      fi
+    fi
+  done
+}
+
+# Apply factor type fixes in idf file
+applyFixes $exp automatic_fixes_properties.txt idf.txt
+if [ $? -ne 0 ]; then
+  echo "ERROR: Applying factor type fixes in idf file for $exp failed" >&2
+  return 1
+fi
+
+# Apply factor/sample characteristic type fixes to the condensed-sdrf file
+applyFixes $exp automatic_fixes_properties.txt condensed-sdrf.tsv
+if [ $? -ne 0 ]; then
+  echo "ERROR: Applying sample characteristic/factor types fixes in sdrf file for $exp failed" >&2
+  return 1
+fi
+# Apply sample characteristic/factor value fixes to the condensed-sdrf file
+applyFixes $exp automatic_fixes_values.txt condensed-sdrf.tsv
+if [ $? -ne 0 ]; then
+  echo "ERROR: Applying sample characteristic/factor value fixes in sdrf file for $exp failed" >&2
+  return 1
+fi

atlas-experiment-metadata-test.bats

@@ -92,3 +92,6 @@ setup() {
     [ -f "$test_unmelted_sdrf" ]
 }
 
+@test "Test get experiment type from XML" {
+    run perl get_experiment_type_from_xml.pl 
+}

get_experiment_type_from_xml.pl

@@ -0,0 +1,14 @@
+#!/usr/bin/env perl
+#
+
+use strict;
+use warnings;
+use 5.10.0;
+
+use Atlas::AtlasConfig::Reader qw( parseAtlasConfig );
+
+my $xmlFilename = shift;
+my $experimentConfig = parseAtlasConfig( $xmlFilename );
+my $experimentType = $experimentConfig->get_atlas_experiment_type;
+
+print $experimentType;

get_magetab_for_experiment.sh

@@ -0,0 +1,62 @@
+#!/usr/bin/env bash
+
+
+usage() { echo "Usage: $0 <accession> path-to-atlas-exps" 1>&2; }
+
+expAcc=$1
+ATLAS_EXPS=$2
+
+if [ -z "${expAcc}" ] || [ -z "${ATLAS_EXPS}" ]; then
+    usage
+    exit 1
+fi
+
+scriptDir=$(cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )
+
+# Get the experiment type from the XML config.
+expType=$(${scriptDir}/get_experiment_type_from_xml.pl $expAcc/$expAcc-configuration.xml)
+if [ $? -ne 0 ]; then
+    echo "ERROR: failed to get $expAcc experiment type from XML config. Cannot generate condensed SDRF."
+    exit 1
+fi
+
+# Now generate condensed sdrf containing ontology mappings from Zooma. This
+# will also copy IDF from ArrayExpress load directory (using "-i" option).
+# If this is a baseline experiment, pass the factors XML filename as well to ensure factors match in condensed SDRF.
+if [[ $expType == *baseline ]]; then
+
+    ${scriptDir}/condense_sdrf.pl -e $expAcc -f $expAcc/$expAcc-factors.xml -z -i -o $expAcc
+    if [ $? -ne 0 ]; then
+        echo "ERROR: Failed to generate $expAcc/${expAcc}.condensed-sdrf.tsv with Zooma mappings, trying without..."
+        ${scriptDir}/condense_sdrf.pl -e $expAcc -f $expAcc/$expAcc-factors.xml -i -o $expAcc
+    fi
+    if [ $? -ne 0 ]; then
+        echo "ERROR: Failed to generate $expAcc/${expAcc}.condensed-sdrf.tsv"
+        return 1
+    fi
+else
+
+    ${scriptDir}/condense_sdrf.pl -e $expAcc -z -i -o $expAcc
+    if [ $? -ne 0 ]; then
+        echo "ERROR: Failed to generate $expAcc/${expAcc}.condensed-sdrf.tsv with Zooma mappings, trying without..."
+        ${scriptDir}/condense_sdrf.pl -e $expAcc -i -o $expAcc
+    fi
+    if [ $? -ne 0 ]; then
+        echo "ERROR: Failed to generate $expAcc/${expAcc}.condensed-sdrf.tsv"
+        return 1
+    fi
+fi
+
+if [ ! -s "$expAcc/${expAcc}.condensed-sdrf.tsv" ]; then
+echo "ERROR: Failed to generate $expAcc/${expAcc}.condensed-sdrf.tsv"
+return 1
+fi
+
+applyAllFixesForExperiment $expAcc
+if [ $? -ne 0 ]; then
+echo "ERROR: Applying fixes for experiment $e failed" >&2
+return 1
+fi
+
+rm -rf $expAcc/$expAcc-zoomifications-log.tsv
+popd