From 8e77e5772688885680620e8e1005ab53e0825662 Mon Sep 17 00:00:00 2001
From: Nathan Gaddis <ngaddis@rti.org>
Date: Wed, 1 May 2024 22:12:59 +0000
Subject: [PATCH] Updates to t1dgrs2 pipeline

---
 .../v1.0/filter_gvcfs_by_consent.sh           |  9 +++++---
 t1dgrs2_pipeline/v1.0/merge_results_files.sh  | 23 +++++++++++++++----
 .../v1.0/t1dgrs2_pipeline_export.json         | 12 +++++++---
 .../v1.0/t1dgrs2_pipeline_sample.json         |  2 +-
 t1dgrs2_pipeline/v1.0/t1dgrs2_tasks.json      | 17 ++++++++++----
 5 files changed, 47 insertions(+), 16 deletions(-)

diff --git a/t1dgrs2_pipeline/v1.0/filter_gvcfs_by_consent.sh b/t1dgrs2_pipeline/v1.0/filter_gvcfs_by_consent.sh
index 5731670..7324778 100644
--- a/t1dgrs2_pipeline/v1.0/filter_gvcfs_by_consent.sh
+++ b/t1dgrs2_pipeline/v1.0/filter_gvcfs_by_consent.sh
@@ -136,9 +136,12 @@ echo "Successfully moved all consented files to gvcfs folder"
 #Move remaining gz files that don't have consent to the alternate folder
 remaining_files=$(ls $imported_gvcfs_dir*.gz |wc -l | cut -d ' ' -f1)
 
-echo "Moving the remaining $remaining_files gvcf files. Along with the tbi files. Will remove them later"
-
-mv $imported_gvcfs_dir*.gz* $nonconsented_gvcfs_dir
+echo "Moving the remaining $remaining_files gvcf files."
+for file in $(ls $imported_gvcfs_dir*.gvcf.gz); do
+	mv $file $nonconsented_gvcfs_dir
+	file_name=$(basename "$file")
+	echo "Moved $file_name to $nonconsented_gvcfs_dir"
+done
 echo "Successfully moved nonconsented gvcfs to nonconsent-gvcfs folder"
 
 ################################
diff --git a/t1dgrs2_pipeline/v1.0/merge_results_files.sh b/t1dgrs2_pipeline/v1.0/merge_results_files.sh
index 2b37681..1556d06 100644
--- a/t1dgrs2_pipeline/v1.0/merge_results_files.sh
+++ b/t1dgrs2_pipeline/v1.0/merge_results_files.sh
@@ -1,7 +1,8 @@
 #!/bin/bash
 
 sample_results_dir=""
-output_file=""
+output_result_file=""
+output_control_file=""
 
 while [ "$1" != "" ]; 
 do
@@ -9,8 +10,11 @@ do
 		--sample_results_dir )		shift
 									sample_results_dir=$1
 									;;
-		--output_file )			    shift
-									output_file=$1
+		--output_result_file )		shift
+									output_result_file=$1
+									;;
+		--output_control_file )		shift
+									output_control_file=$1
 									;;
 	esac
 	shift
@@ -20,10 +24,19 @@ sample_results_dir=$(echo $sample_results_dir | perl -ne 'chomp; if (substr($_,
 mkdir -p $sample_results_dir
 
 first_file=true
-for file in $(ls $sample_results_dir*/*_for_export.tsv ); do
+for file in $(ls $sample_results_dir*/*_for_export.tsv | grep -v qc ); do
+    if [ "$first_file" = true ] ; then
+        head -n 1 $file
+        first_file=false
+    fi
+    tail -n +2 $file
+done > $output_result_file
+
+first_file=true
+for file in $(ls $sample_results_dir*/*_for_export.tsv | grep qc ); do
     if [ "$first_file" = true ] ; then
         head -n 1 $file
         first_file=false
     fi
     tail -n +2 $file
-done > $output_file
+done > $output_control_file
diff --git a/t1dgrs2_pipeline/v1.0/t1dgrs2_pipeline_export.json b/t1dgrs2_pipeline/v1.0/t1dgrs2_pipeline_export.json
index 902541a..0b4359e 100644
--- a/t1dgrs2_pipeline/v1.0/t1dgrs2_pipeline_export.json
+++ b/t1dgrs2_pipeline/v1.0/t1dgrs2_pipeline_export.json
@@ -16,7 +16,12 @@
             "default": "",
             "required": true
         },
-        "output_file": {
+        "output_result_file": {
+            "type": "string",
+            "default": "",
+            "required": true
+        },
+        "output_contorl_file": {
             "type": "string",
             "default": "",
             "required": true
@@ -51,7 +56,8 @@
     "entry_point": "merge_results",
     "entry_point_inputs": {
         "sample_results_dir": "<inputs.results_dir>",
-        "output_file": "<inputs.output_file>"
+        "output_result_file": "<inputs.output_result_file>",
+        "output_contorl_file": "<inputs.output_contorl_file>"
     },
     "pipeline": {
         "merge_results": {
@@ -63,7 +69,7 @@
                         "sftp_server": "<inputs.sftp_server>",
                         "username": "<inputs.username>",
                         "password": "<inputs.password>",
-                        "results_file": "<inputs.output_file>",
+                        "results_file": "<inputs.output_result_file>",
                         "target_dir": "<inputs.target_dir>"
                     }
                 },
diff --git a/t1dgrs2_pipeline/v1.0/t1dgrs2_pipeline_sample.json b/t1dgrs2_pipeline/v1.0/t1dgrs2_pipeline_sample.json
index 9b8ecac..61da026 100644
--- a/t1dgrs2_pipeline/v1.0/t1dgrs2_pipeline_sample.json
+++ b/t1dgrs2_pipeline/v1.0/t1dgrs2_pipeline_sample.json
@@ -142,7 +142,7 @@
                         "missing_file": "<steps.extract_gvcf_variants.outputs.missing>",
                         "sample_id": "<inputs.sample_id>",
                         "genedx_manifest": "<inputs.genedx_manifest>",
-                        "output_file": "<inputs.output_dir><inputs.sample_id>_for_export.tsv"
+                        "output_file": "<inputs.output_dir><inputs.sample_id>_for_export.csv"
                     }
                 },
                 "returncode_1": {
diff --git a/t1dgrs2_pipeline/v1.0/t1dgrs2_tasks.json b/t1dgrs2_pipeline/v1.0/t1dgrs2_tasks.json
index ac3bbb3..8e66ff7 100644
--- a/t1dgrs2_pipeline/v1.0/t1dgrs2_tasks.json
+++ b/t1dgrs2_pipeline/v1.0/t1dgrs2_tasks.json
@@ -489,7 +489,10 @@
             "sample_results_dir": {
                 "type": "dir"
             },
-            "output_file": {
+            "output_result_file": {
+                "type": "str"
+            },
+            "output_control_file": {
                 "type": "str"
             }
         },
@@ -497,13 +500,19 @@
             "/opt/merge_results_files.sh",
             "--sample_results_dir",
             "<inputs.sample_results_dir>",
-            "--output_file",
-            "<inputs.output_file>"
+            "--output_result_file",
+            "<inputs.output_result_file>",
+            "--output_control_file",
+            "<inputs.output_control_file>"
         ],
         "outputs": {
             "results_file": {
                 "type": "file",
-                "value": "<inputs.output_file>"
+                "value": "<inputs.output_result_file>"
+            },
+            "control_file": {
+                "type": "file",
+                "value": "<inputs.output_control_file>"
             },
             "log": {
                 "type": "file",