Bug fixes for t1dgrs2 v2 pipeline

RTIInternational · Sep 20, 2024 · d210a12 · d210a12
1 parent 003180b
commit d210a12
Show file tree

Hide file tree

Showing 4 changed files with 77 additions and 27 deletions.
diff --git a/t1dgrs2_pipeline/v2.0/pipeline_config/entrypoint_extract_gvcf_variants.json b/t1dgrs2_pipeline/v2.0/pipeline_config/entrypoint_extract_gvcf_variants.json
@@ -11,18 +11,18 @@
             "default": "",
             "required": true
         },
-        "gvcf": {
-            "type": "file",
+        "sample_id": {
+            "type": "str",
             "default": "",
             "required": true
         },
-        "variant_list": {
+        "gvcf": {
             "type": "file",
             "default": "",
             "required": true
         },
-        "sample_id": {
-            "type": "str",
+        "variant_list": {
+            "type": "file",
             "default": "",
             "required": true
         },
@@ -69,14 +69,14 @@
         "output_dir": "<inputs.output_dir>extract_gvcf_variants",
         "sample_id": "<inputs.sample_id>",
         "gvcf": "<inputs.gvcf>",
-        "out_prefix": "<inputs.output_dir>extract_gvcf_variants/<inputs.sample_id>",
         "variant_list": "<inputs.variant_list>",
+        "hla_variants_file": "<inputs.hla_variants_file>",
+        "non_hla_variants_file": "<inputs.non_hla_variants_file>",
+        "out_prefix": "<inputs.output_dir>extract_gvcf_variants/<inputs.sample_id>",
         "pass_only": "<inputs.pass_only>",
         "filter_by_gq": "<inputs.filter_by_gq>",
         "hom_gq_threshold": "<inputs.hom_gq_threshold>",
-        "het_gq_threshold": "<inputs.het_gq_threshold>",
-        "hla_variants_file": "<inputs.hla_variants_file>",
-        "non_hla_variants_file": "<inputs.non_hla_variants_file>"
+        "het_gq_threshold": "<inputs.het_gq_threshold>"
     },
     "pipeline": {
         "extract_gvcf_variants": {

diff --git a/t1dgrs2_pipeline/v2.0/pipeline_config/t1dgrs2_tasks.json b/t1dgrs2_pipeline/v2.0/pipeline_config/t1dgrs2_tasks.json
@@ -46,7 +46,7 @@
             "--target_dir",
             "<inputs.target_dir>",
             "--manifest_dir",
-            "<inputs.rti_manifest_dir>",
+            "<inputs.manifest_dir>",
             "--downloaded_samples",
             "<inputs.downloaded_samples>",
             "--download_limit",
@@ -136,6 +136,12 @@
             "variant_list": {
                 "type": "file"
             },
+            "hla_variants_file": {
+                "type": "file"
+            },
+            "non_hla_variants_file": {
+                "type": "file"
+            },
             "out_prefix": {
                 "type": "str"
             },
@@ -154,22 +160,22 @@
             "het_gq_threshold": {
                 "type": "int",
                 "default": 48
-            },
-            "hla_variants_file": {
-                "type": "file"
-            },
-            "non_hla_variants_file": {
-                "type": "file"
             }
         },
         "cmd": [
             "/opt/extract_gvcf_variants.pl",
+            "--sample_id",
+            "<inputs.sample_id>",
             "--gvcf",
             "<inputs.gvcf>",
-            "--out_prefix",
-            "<inputs.out_prefix>",
             "--variant_list",
             "<inputs.variant_list>",
+            "--hla_variants_file",
+            "<inputs.hla_variants_file>",
+            "--non_hla_variants_file",
+            "<inputs.non_hla_variants_file>",
+            "--out_prefix",
+            "<inputs.out_prefix>",
             "--pass_only",
             "<inputs.pass_only>",
             "--filter_by_gq",

diff --git a/t1dgrs2_pipeline/v2.0/scripts/batch_extract_gvcf_variants.py b/t1dgrs2_pipeline/v2.0/scripts/batch_extract_gvcf_variants.py
@@ -30,6 +30,18 @@
     help = 'List of variants to extract',
     type = str
 )
+parser.add_argument(
+    '--hla_variants_file',
+    required = True,
+    help = 'File listing HLA variants used for DQ imputation',
+    type = str
+)
+parser.add_argument(
+    '--non_hla_variants_file',
+    required = True,
+    help = 'File listing HLA variants not used for DQ imputation',
+    type = str
+)
 parser.add_argument(
     '--argo_api_url',
     required = False,
@@ -44,6 +56,34 @@
     default = 50,
     type = int
 )
+parser.add_argument(
+    '--pass_only',
+    required = False,
+    help = 'Limit extraction to variants designated PASS',
+    default = 0,
+    type = int
+)
+parser.add_argument(
+    '--filter_by_gq',
+    required = False,
+    help = 'Filter variatns by GQ',
+    default = 0,
+    type = int
+)
+parser.add_argument(
+    '--hom_gq_threshold',
+    required = False,
+    help = 'GQ threshold for homozygous variants',
+    default = 99,
+    type = int
+)
+parser.add_argument(
+    '--het_gq_threshold',
+    required = False,
+    help = 'GQ threshold for heterozygous variants',
+    default = 48,
+    type = int
+)
 args = parser.parse_args()
 
 # Function to get the number of running workflows
@@ -100,13 +140,15 @@ def get_running_workflows():
     wf_arguments = {
         "working_dir": sample_working_dir,
         "output_dir": sample_output_dir,
+        "sample_id": file_id,
         "gvcf": path,
         "variant_list": args.variant_list,
-        "sample_id": file_id,
-        "pass_only": 0,
-        "filter_by_gq": 0,
-        "hom_gq_threshold": 99,
-        "het_gq_threshold": 48
+        "hla_variants_file": args.hla_variants_file,
+        "non_hla_variants_file": args.non_hla_variants_file,
+        "pass_only": args.pass_only,
+        "filter_by_gq": args.filter_by_gq,
+        "hom_gq_threshold": args.hom_gq_threshold,
+        "het_gq_threshold": args.het_gq_threshold
     }
     file_wf_arguments = working_dir + file_id + '.json'
     with open(file_wf_arguments, 'w', encoding='utf-8') as f:

diff --git a/t1dgrs2_pipeline/v2.0/scripts/extract_gvcf_variants.pl b/t1dgrs2_pipeline/v2.0/scripts/extract_gvcf_variants.pl
@@ -11,8 +11,10 @@
 
 my $sample_id = '';
 my $gvcf = '';
-my $out_prefix;
 my $variant_list = '';
+my $hla_variants_file = '';
+my $non_hla_variants_file = '';
+my $out_prefix;
 my $pass_only = 0;
 my $filter_by_gq = 0;
 my $hom_gq_threshold = 99;
@@ -23,14 +25,14 @@
 GetOptions (
     'sample_id=s' => \$sample_id,
     'gvcf=s' => \$gvcf,
-    'out_prefix=s' => \$out_prefix,
     'variant_list=s' => \$variant_list,
+    'hla_variants_file' => \$hla_variants_file,
+    'non_hla_variants_file' => \$non_hla_variants_file
+    'out_prefix=s' => \$out_prefix,
     'pass_only:i' => \$pass_only,
     'filter_by_gq:i' => \$filter_by_gq,
     'hom_gq_threshold:i' => \$hom_gq_threshold,
     'het_gq_threshold:i' => \$het_gq_threshold,
-    'hla_variants_file' => \$hla_variants_file,
-    'non_hla_variants_file' => \$non_hla_variants_file
 ) or die("Invalid options");
 
 sub flip {