As a Developer, provide a way to remove all automatic annotations fro…

…m analyses, so they can be re-annotated with the correct information (#99) * Adding the remove annotation and annotation queueing scripts * Adding documentation and run feedback * Updated the analysis annotation script to give relevant output * Fixed the annotation config that broke annotation * Linting should be passing * Removed debugging print statements * Added a note to the annotate-all-existing-analyses.sh that it may need to be run more than once * Removed a space
uab-cgds-worthey · Jul 5, 2023 · 068fffb · 068fffb
1 parent bfbe92e
commit 068fffb
Show file tree

Hide file tree

Showing 3 changed files with 145 additions and 8 deletions.
diff --git a/etc/api/annotate-all-existing-analyses.sh b/etc/api/annotate-all-existing-analyses.sh
@@ -0,0 +1,77 @@
+#! /bin/bash
+# ./annotate-all-existing-analyses.sh
+
+usage() {
+  echo " "
+  echo "usage: $0"
+  echo " "
+  echo " -h Prints usage"
+  echo " "
+  echo "Kicks off annotation jobs for all the existing analyses in Rosalution"
+  echo " "
+  echo "To run the annotations, please log in to Rosalution to retrieve the"
+  echo "Client ID and Client Secret credentials. These can be found by"
+  echo "clicking on your username or going to: <rosalution url>/rosalution/account"
+  echo " "
+  echo "Note: This script may need to be run multiple times due to the Rosalution"
+  echo "annotation system not being built for large sets of annotations queued."
+  echo " "
+  echo "Please install jq for this script to work. https://stedolan.github.io/jq/"
+  echo " "
+  exit
+}
+
+while getopts ":h" opt; do
+  case $opt in
+    h) usage;;
+    \?) echo "Invalid option -$OPTARG" && exit 127;;
+  esac
+done
+
+if ! jq --version &> /dev/null
+then
+    echo "Error: jq could not be found. Exiting."
+    usage
+fi
+
+echo "Please enter your Client Id";
+read -r CLIENT_ID;
+
+echo "Please enter your Client Secret";
+read -r -s CLIENT_SECRET;
+
+if [ -z "${CLIENT_ID}" ] || [ -z "${CLIENT_SECRET}" ]; then
+  echo " "
+  echo "Please enter required credentials."
+  usage
+fi
+
+echo "Fetching valid authentication token..."
+
+AUTH_TOKEN=$(curl -s -X 'POST' \
+  "http://local.rosalution.cgds/rosalution/api/auth/token" \
+  -H "accept: application/json" \
+  -H "Content-Type: application/x-www-form-urlencoded" \
+  -d "grant_type=&scope=&client_id=$CLIENT_ID&client_secret=$CLIENT_SECRET" | jq -r '.access_token')
+
+echo "Fetching existing analyses in Rosalution..."
+
+RESPONSE=$(curl -s -X "GET" \
+  "http://local.rosalution.cgds/rosalution/api/analysis/" \
+  -H "accept: application/json" \
+  -H "Authorization: Bearer $AUTH_TOKEN")
+
+ANALYSES=()
+while IFS='' read -r line; do ANALYSES+=("$line"); done < <(echo "$RESPONSE" | jq -c '[.[].name]')
+
+echo "${ANALYSES[@]}" | jq -r '.[]' | while read -r ANALYSIS; do
+  echo "Starting annotations for analysis $ANALYSIS..."
+  curl -s -X "POST" \
+      "http://local.rosalution.cgds/rosalution/api/annotate/$ANALYSIS" \
+      -H "accept: application/json" \
+      -H "Authorization: Bearer $AUTH_TOKEN" \
+      > /dev/null
+  sleep 5
+done
+
+echo "Done. Exiting."
diff --git a/etc/fixtures/initial-seed/annotations-config.json b/etc/fixtures/initial-seed/annotations-config.json
@@ -325,9 +325,9 @@
   {
     "data_set": "Rat_Alliance_Genome_url",
     "data_source": "Rosalution",
-    "annotation_source_type": "forge",
     "genomic_unit_type": "gene",
-    "url": "https://www.alliancegenome.org/gene/{Rat Gene Identifier}",
+    "annotation_source_type": "forge",
+    "base_string": "https://www.alliancegenome.org/gene/{Rat Gene Identifier}",
     "attribute": "{ \"Rat_Alliance_Genome_url\": .Rat_Alliance_Genome_url }",
     "dependencies": ["Rat Gene Identifier"]
   },
@@ -343,9 +343,9 @@
   {
     "data_set": "Mouse_Alliance_Genome_url",
     "data_source": "Rosalution",
-    "annotation_source_type": "forge",
     "genomic_unit_type": "gene",
-    "url": "https://www.alliancegenome.org/gene/{Mouse Gene Identifier}",
+    "annotation_source_type": "forge",
+    "base_string": "https://www.alliancegenome.org/gene/{Mouse Gene Identifier}",
     "attribute": "{ \"Mouse_Alliance_Genome_url\": .Mouse_Alliance_Genome_url }",
     "dependencies": ["Mouse Gene Identifier"]
   },
@@ -361,9 +361,9 @@
   {
     "data_set": "Zebrafish_Alliance_Genome_url",
     "data_source": "Rosalution",
-    "annotation_source_type": "forge",
     "genomic_unit_type": "gene",
-    "url": "https://www.alliancegenome.org/gene/{Zebrafish Gene Identifier}",
+    "annotation_source_type": "forge",
+    "base_string": "https://www.alliancegenome.org/gene/{Zebrafish Gene Identifier}",
     "attribute": "{ \"Zebrafish_Alliance_Genome_url\": .Zebrafish_Alliance_Genome_url }",
     "dependencies": ["Zebrafish Gene Identifier"]
   },
@@ -379,9 +379,9 @@
   {
     "data_set": "C-Elegens_Alliance_Genome_url",
     "data_source": "Rosalution",
-    "annotation_source_type": "forge",
     "genomic_unit_type": "gene",
-    "url": "https://www.alliancegenome.org/gene/{C-Elegens Gene Identifier}",
+    "annotation_source_type": "forge",
+    "base_string": "https://www.alliancegenome.org/gene/{C-Elegens Gene Identifier}",
     "attribute": "{ \"C-Elegens_Alliance_Genome_url\": .C-Elegens_Alliance_Genome_url }",
     "dependencies": ["C-Elegens Gene Identifier"]
   },

diff --git a/etc/fixtures/remove-automatic-annotations.js b/etc/fixtures/remove-automatic-annotations.js
@@ -0,0 +1,60 @@
+const usage = `
+Script usage for 'remove-automatic-annotations.js':
+
+mongosh /tmp/fixtures/remove-automatic-annotations.js
+    Script Options:
+        help            If true, prints this message.
+        databaseName    Database name to use - default: rosalution_db
+    
+    For mongosh connection and authentication usage, please run: mongosh help
+
+    Examples:
+        mongosh --host localhost --port 27017 /tmp/fixtures/remove-automatic-annotations.js
+        mongosh --host localhost --port 27017 --eval "help=true;databaseName='your_db_name'" /tmp/fixtures/remove-automatic-annotations.js 
+`
+
+if(help == true) {
+    print(usage);
+    quit(1);
+}
+
+if(typeof databaseName == 'undefined')
+    databaseName = 'rosalution_db';
+else if(typeof databaseName !== 'string') {
+    print("databaseName must be a string");
+    quit(1);
+}
+
+db = db.getSiblingDB(databaseName);
+
+console.log(`Removing non-manual annotations from ${databaseName}...`);
+
+try {
+    const genomic_units = db.genomic_units.find();
+
+    if(genomic_units.size() == 0)
+        throw new Error(`No genomic units found in ${databaseName}. Aborting.`)
+
+    const newUnits = [];
+    genomic_units.forEach(unit => {
+        unit.annotations.forEach(annotation => {
+            for(const [key, value] of Object.entries(annotation)) {
+                annotation[key] = annotation[key].filter(object => object['data_source'] == 'rosalution-manual')
+                if(annotation[key].length < 1)
+                    delete annotation[key];
+            }
+        })
+
+        newUnits.push(unit);
+    });
+
+    newUnits.forEach(unit => {unit.annotations = unit.annotations.filter(object => Object.keys(object).length !== 0)});
+    newUnits.forEach(unit => {db.genomic_units.updateOne({'_id': unit._id}, {'$set': unit})});
+
+} catch (err) {
+    console.log(err.stack);
+    console.log(usage);
+    quit(1);
+}
+
+console.log(`Annotation removal complete.`);