File of callset samples -> samples marked as 'withdrawn' in GVS [VS-436]

.dockstore.yml

@@ -179,6 +179,7 @@ workflows:
        branches:
          - master
          - ah_var_store
+         - rsa_vs_436_withdrawn_wdl
    - name: GvsUnified
      subclass: WDL
      primaryDescriptorPath: /scripts/variantstore/wdl/GvsUnified.wdl

scripts/variantstore/wdl/GvsWithdrawSamples.wdl

@@ -6,14 +6,19 @@ workflow GvsWithdrawSamples {
     String dataset_name
     String project_id
 
-    Array[String] sample_names
+    String sample_name_column_name_in_file
+    File sample_names_to_include_file
+    # should be in the format "2022-01-01 00:00:00 UTC"
+    String withdrawn_timestamp
   }
 
   call WithdrawSamples {
     input:
       project_id = project_id,
       dataset_name = dataset_name,
-      sample_names = sample_names
+      sample_name_column_name_in_file = sample_name_column_name_in_file,
+      sample_names_to_include_file = sample_names_to_include_file,
+      withdrawn_timestamp = withdrawn_timestamp
   }
 
   output {
@@ -26,40 +31,50 @@ task WithdrawSamples {
     String project_id
     String dataset_name
 
-    Array[String] sample_names
+    String sample_name_column_name_in_file
+    File sample_names_to_include_file
+    # should be in the format "2022-01-01 00:00:00 UTC"
+    String withdrawn_timestamp
   }
 
   meta {
-    description: "Withdraw Samples from GVS by marking them as 'withdrawn' in the sample_info table"
+    description: "Given a list of samples for a callset, withdraw samples from GVS that are not included by marking them as 'withdrawn' in the sample_info table with a passed in timestamp."
     # Might not be strictly necessary to make this volatile, but just in case:
     volatile: true
   }
 
   command <<<
-    set -e
-    set -x
+    set -o errexit -o nounset -o xtrace -o pipefail
+    echo "project_id = ~{project_id}" > ~/.bigqueryrc
+
+    # get just the sample_name values from sample_names_to_include_file based on the
+    # sample_name_column_name_in_file into sample_names.tsv
+    col_num=$(head -n1 ~{sample_names_to_include_file} | tr '\t' '\n' | grep -Fxn '~{sample_name_column_name_in_file}' | cut -f1 -d:)
+    awk "{print \$$col_num}" ~{sample_names_to_include_file} | sed 1d > sample_names.tsv
 
-    # make sure that sample names were actually passed, warn and exit if empty
-    num_samples=~{length(sample_names)}
+    # make sure that we end up with some samples to make the temp table, warn and exit if empty
+    num_samples=$(cat sample_names.tsv | wc -l)
     if [ $num_samples -eq 0 ]; then
-      echo "No sample names passed. Exiting"
+      echo "No sample names for callset produced. Exiting"
       exit 0
     fi
 
-    echo "project_id = ~{project_id}" > ~/.bigqueryrc
+    # create the temp table (expires in 1 day)
+    bq --project_id=~{project_id} mk --expiration=86400 ~{dataset_name}.current_callset_samples "sample_name:STRING"
+    # populate the temp table
+    bq load --project_id=~{project_id} -F "tab" ~{dataset_name}.current_callset_samples sample_names.tsv
 
-    # perform actual update
+    # join on the temp table to figure out which samples should be marked as withdrawn
     bq --project_id=~{project_id} query --format=csv --use_legacy_sql=false \
-      'UPDATE `~{dataset_name}.sample_info` SET withdrawn = CURRENT_TIMESTAMP() WHERE sample_name IN ("~{sep='\", \"' sample_names}")' > log_message.txt;
+      "UPDATE \`~{dataset_name}.sample_info\` AS samples SET withdrawn = '~{withdrawn_timestamp}' \
+        WHERE NOT EXISTS \
+        (SELECT * \
+        FROM \`~{project_id}.~{dataset_name}.current_callset_samples\` AS callset \
+        WHERE \
+        samples.sample_name = callset.sample_name \
+        AND NOT samples.is_control)" > log_message.txt
 
     cat log_message.txt | sed -e 's/Number of affected rows: //' > rows_updated.txt
-    typeset -i rows_updated=$(cat rows_updated.txt)
-
-    if [ $num_samples -ne $rows_updated ]; then
-      echo "Error: Expected to update $num_samples rows - but only updated $rows_updated."
-      exit 1
-    fi
-
   >>>
   runtime {
     docker: "us.gcr.io/broad-gatk/gatk:4.2.5.0"