Merge pull request #98 from PacificBiosciences/feature/deepvariant-1.6.0

feat: Update to DeepVariant v1.6.0
PacificBiosciences · Dec 15, 2023 · f92576b · f92576b
2 parents 725861d + fc8b47f
commit f92576b
Show file tree

Hide file tree

Showing 11 changed files with 25 additions and 25 deletions.
diff --git a/README.md b/README.md
@@ -172,8 +172,8 @@ These files are hosted publicly in each of the cloud backends; see `backends/${b
 
 | Type | Name | Description | Notes |
 | :- | :- | :- | :- |
-| String? | deepvariant_version | Version of deepvariant to use \["1.5.0"\] | |
-| [DeepVariantModel](https://github.com/PacificBiosciences/wdl-common/blob/main/wdl/structs.wdl)? | deepvariant_model | Optional alternate DeepVariant model file to use | |
+| String? | deepvariant_version | Version of deepvariant to use \["1.6.0"\] | |
+ | File? | custom_deepvariant_model_tar | Optional alternate DeepVariant [custom model](https://github.com/PacificBiosciences/wdl-common/blob/8d868818b62345fdb64ab104238d406d65334b3e/wdl/workflows/deepvariant/README.md) to use | |
 | Int? | pbsv_call_mem_gb | Optionally set RAM (GB) for pbsv_call during cohort analysis | |
 | Int? | glnexus_mem_gb | Optionally set RAM (GB) for GLnexus during cohort analysis | |
 | Boolean? | run_tertiary_analysis | Run the optional tertiary analysis steps \[`false`\] | \[`true`, `false`\] |
@@ -256,7 +256,7 @@ The Docker image used by a particular step of the workflow can be identified by
 | Image | Major tool versions | Links |
 | :- | :- | :- |
 | bcftools | <ul><li>[bcftools 1.14](https://github.com/samtools/bcftools/releases/tag/1.14)</li></ul> | [Dockerfile](https://github.com/PacificBiosciences/wdl-dockerfiles/tree/3560fcc5a84e044067cea9c9a7669cfc2659178e/docker/bcftools) |
-| deepvariant | User-defined; default is version [1.5.0](https://github.com/google/deepvariant/releases/tag/v1.5.0) | [DeepVariant GitHub](https://github.com/google/deepvariant) |
+| deepvariant | User-defined; default is version [1.6.0](https://github.com/google/deepvariant/releases/tag/v1.6.0) | [DeepVariant GitHub](https://github.com/google/deepvariant) |
 | glnexus | <ul><li>[glnexus v1.4.3](https://github.com/dnanexus-rnd/GLnexus/releases/tag/v1.4.3)</li></ul> | [GLnexus GitHub](https://github.com/dnanexus-rnd/GLnexus) |
 | hificnv | <ul><li>[HiFiCNV v0.1.7](https://github.com/PacificBiosciences/HiFiCNV/releases/tag/v0.1.7)</li><li>[bcftools 1.16](https://github.com/samtools/bcftools/releases/tag/1.16)</li></ul> | [Dockerfile](https://github.com/PacificBiosciences/wdl-dockerfiles/tree/0b0fbe939648087e9fdea4497ae08dc76538ebf0/docker/hificnv) |
 | hiphase | <ul><li>[HiPhase 1.0.0](https://github.com/PacificBiosciences/HiPhase/releases/tag/v1.0.0)</li><li>[samtools 1.18](https://github.com/samtools/samtools/releases/tag/1.18)</li><li>[bcftools 1.18](https://github.com/samtools/bcftools/releases/tag/1.18)</li></ul> | [Dockerfile](https://github.com/PacificBiosciences/wdl-dockerfiles/tree/d26db6204409dfeff56e169cdba0cc14bc272f15/docker/hiphase) |

diff --git a/backends/aws/inputs.aws.json b/backends/aws/inputs.aws.json
@@ -64,7 +64,7 @@
     "clinvar_lookup": "s3://dnastack-resources/dataset/slivar/clinvar_gene_desc.20221214T183140.txt"
   },
   "humanwgs.deepvariant_version": "String (optional)",
-  "humanwgs.deepvariant_model": "WomCompositeType {\n model -> WomCompositeType {\n data -> File\ndata_index -> File \n}\nmetadata -> File \n}? (optional)",
+  "humanwgs.custom_deepvariant_model_tar": "File (optional)",
   "humanwgs.pbsv_call_mem_gb": "Int (optional, default = if N<=3: 64 else 96)",
   "humanwgs.glnexus_mem_gb": "Int (optional, default = 30)",
   "humanwgs.run_tertiary_analysis": "Boolean (optional, default = false)",

diff --git a/backends/azure/inputs.azure.json b/backends/azure/inputs.azure.json
@@ -64,7 +64,7 @@
     "clinvar_lookup": "/datasetpbrarediseases/dataset/slivar/clinvar_gene_desc.20221214T183140.txt"
   },
   "humanwgs.deepvariant_version": "String (optional)",
-  "humanwgs.deepvariant_model": "WomCompositeType {\n model -> WomCompositeType {\n data -> File\ndata_index -> File \n}\nmetadata -> File \n}? (optional)",
+  "humanwgs.custom_deepvariant_model_tar": "File (optional)",
   "humanwgs.pbsv_call_mem_gb": "Int (optional, default = if N<=3: 64 else 96)",
   "humanwgs.glnexus_mem_gb": "Int (optional, default = 30)",
   "humanwgs.run_tertiary_analysis": "Boolean (optional, default = false)",

diff --git a/backends/gcp/inputs.gcp.json b/backends/gcp/inputs.gcp.json
@@ -64,7 +64,7 @@
     "clinvar_lookup": "gs://pacbio-wdl/dataset/slivar/clinvar_gene_desc.20221214T183140.txt"
   },
   "humanwgs.deepvariant_version": "String (optional)",
-  "humanwgs.deepvariant_model": "WomCompositeType {\n model -> WomCompositeType {\n data -> File\ndata_index -> File \n}\nmetadata -> File \n}? (optional)",
+  "humanwgs.custom_deepvariant_model_tar": "File (optional)",
   "humanwgs.pbsv_call_mem_gb": "Int (optional, default = if N<=3: 64 else 96)",
   "humanwgs.glnexus_mem_gb": "Int (optional, default = 30)",
   "humanwgs.run_tertiary_analysis": "Boolean (optional, default = false)",

diff --git a/backends/hpc/inputs.hpc.json b/backends/hpc/inputs.hpc.json
@@ -64,7 +64,7 @@
     "clinvar_lookup": "<local_path_prefix>/dataset/slivar/clinvar_gene_desc.20221214T183140.txt"
   },
   "humanwgs.deepvariant_version": "String (optional)",
-  "humanwgs.deepvariant_model": "WomCompositeType {\n model -> WomCompositeType {\n data -> File\ndata_index -> File \n}\nmetadata -> File \n}? (optional)",
+  "humanwgs.custom_deepvariant_model_tar": "File (optional)",
   "humanwgs.pbsv_call_mem_gb": "Int (optional, default = if N<=3: 64 else 96)",
   "humanwgs.glnexus_mem_gb": "Int (optional, default = 30)",
   "humanwgs.run_tertiary_analysis": "Boolean (optional, default = false)",

diff --git a/wdl-ci.config.json b/wdl-ci.config.json
@@ -570,7 +570,7 @@
       "tasks": {
         "deepvariant_make_examples": {
           "key": "deepvariant_make_examples",
-          "digest": "35kzpf37semcoxs7frzvhjrc4zvwoyan",
+          "digest": "3jfeho4suf23enopqj23cx6ygviftqny",
           "tests": [
             {
               "inputs": {
@@ -598,7 +598,7 @@
                 "task_start_index": 0,
                 "tasks_per_shard": 8,
                 "total_deepvariant_tasks": 64,
-                "deepvariant_version": "1.5.0",
+                "deepvariant_version": "1.6.0",
                 "runtime_attributes": "${default_runtime_attributes}"
               },
               "output_tests": {
@@ -620,7 +620,7 @@
         },
         "deepvariant_call_variants": {
           "key": "deepvariant_call_variants",
-          "digest": "a6ksi3haiz5pye7p64c67zeeauit7gqf",
+          "digest": "rywffaewvhwakdysxejsrqbz7bqwzt2o",
           "tests": [
             {
               "inputs": {
@@ -637,12 +637,12 @@
                   "${resources_file_path}/deepvariant/${sample_id}.56.example_tfrecords.tar.gz"
                 ],
                 "total_deepvariant_tasks": 64,
-                "deepvariant_version": "1.5.0",
+                "deepvariant_version": "1.6.0",
                 "runtime_attributes": "${default_runtime_attributes}"
               },
               "output_tests": {
-                "tfrecord": {
-                  "value": "${resources_file_path}/deepvariant/${sample_id}.GRCh38.call_variants_output.tfrecord.gz",
+                "tfrecords_tar": {
+                  "value": "${resources_file_path}/deepvariant/${sample_id}.GRCh38.call_variants_output.tar.gz",
                   "test_tasks": [
                     "compare_file_basename",
                     "check_gzip"
@@ -654,12 +654,12 @@
         },
         "deepvariant_postprocess_variants": {
           "key": "deepvariant_postprocess_variants",
-          "digest": "afmkoy2hy4lcyolms7n6lgzuzudr7uxx",
+          "digest": "ey7zqpajaeesvsg372rehhjmkpqld2qx",
           "tests": [
             {
               "inputs": {
                 "sample_id": "${sample_id}",
-                "tfrecord": "${resources_file_path}/deepvariant/${sample_id}.GRCh38.call_variants_output.tfrecord.gz",
+                "tfrecords_tar": "${resources_file_path}/deepvariant/${sample_id}.GRCh38.call_variants_output.tar.gz",
                 "nonvariant_site_tfrecord_tars": [
                   "${resources_file_path}/deepvariant/${sample_id}.0.nonvariant_site_tfrecords.tar.gz",
                   "${resources_file_path}/deepvariant/${sample_id}.8.nonvariant_site_tfrecords.tar.gz",
@@ -674,7 +674,7 @@
                 "reference_index": "${datasets_file_path}/GRCh38/human_GRCh38_no_alt_analysis_set.fasta.fai",
                 "reference_name": "GRCh38",
                 "total_deepvariant_tasks": 64,
-                "deepvariant_version": "1.5.0",
+                "deepvariant_version": "1.6.0",
                 "runtime_attributes": "${default_runtime_attributes}"
               },
               "output_tests": {

diff --git a/workflows/input_template.json b/workflows/input_template.json
@@ -42,7 +42,7 @@
     "clinvar_lookup": "File"
   },
   "humanwgs.deepvariant_version": "String? (optional)",
-  "humanwgs.deepvariant_model": "WomCompositeType {\n model -> WomCompositeType {\n data -> File\ndata_index -> File \n}\nmetadata -> File \n}? (optional)",
+  "humanwgs.custom_deepvariant_model_tar": "File (optional)",
   "humanwgs.pbsv_call_mem_gb": "Int? (optional)",
   "humanwgs.glnexus_mem_gb": "Int? (optional)",
   "humanwgs.run_tertiary_analysis": "Boolean? (optional, default = false)",

diff --git a/workflows/main.wdl b/workflows/main.wdl
@@ -13,8 +13,8 @@ workflow humanwgs {
 		ReferenceData reference
 		SlivarData? slivar_data
 
-		String deepvariant_version = "1.5.0"
-		DeepVariantModel? deepvariant_model
+		String deepvariant_version = "1.6.0"
+		File? custom_deepvariant_model_tar
 
 		Int? pbsv_call_mem_gb
 		Int? glnexus_mem_gb
@@ -48,7 +48,7 @@ workflow humanwgs {
 				sample = sample,
 				reference = reference,
 				deepvariant_version = deepvariant_version,
-				deepvariant_model = deepvariant_model,
+				custom_deepvariant_model_tar = custom_deepvariant_model_tar,
 				default_runtime_attributes = default_runtime_attributes
 		}
 	}

diff --git a/workflows/sample_analysis/inputs.json b/workflows/sample_analysis/inputs.json
@@ -30,7 +30,7 @@
     "population_vcfs": "Array[WomCompositeType {\n data -> File\ndata_index -> File \n}]? (optional)"
   },
   "sample_analysis.deepvariant_version": "String",
-  "sample_analysis.deepvariant_model": "WomCompositeType {\n model -> WomCompositeType {\n data -> File\ndata_index -> File \n}\nmetadata -> File \n}? (optional)",
+  "sample_analysis.custom_deepvariant_model_tar": "File (optional)",
   "sample_analysis.default_runtime_attributes": {
     "preemptible_tries": "Int",
     "max_retries": "Int",

diff --git a/workflows/sample_analysis/sample_analysis.wdl b/workflows/sample_analysis/sample_analysis.wdl
@@ -17,7 +17,7 @@ workflow sample_analysis {
 		ReferenceData reference
 
 		String deepvariant_version
-		DeepVariantModel? deepvariant_model
+		File? custom_deepvariant_model_tar
 
 		RuntimeAttributes default_runtime_attributes
 	}
@@ -56,7 +56,7 @@ workflow sample_analysis {
 			reference_fasta = reference.fasta,
 			reference_name = reference.name,
 			deepvariant_version = deepvariant_version,
-			deepvariant_model = deepvariant_model,
+			custom_deepvariant_model_tar = custom_deepvariant_model_tar,
 			default_runtime_attributes = default_runtime_attributes
 	}
 
@@ -244,7 +244,7 @@ workflow sample_analysis {
 		sample: {help: "Sample information and associated data files"}
 		reference: {help: "Reference genome data"}
 		deepvariant_version: {help: "Version of deepvariant to use"}
-		deepvariant_model: {help: "Optional deepvariant model file to use"}
+		custom_deepvariant_model_tar: {help: "Optional deepvariant model to use"}
 		default_runtime_attributes: {help: "Default RuntimeAttributes; spot if preemptible was set to true, otherwise on_demand"}
 	}
 }

diff --git a/workflows/wdl-common b/workflows/wdl-common
+0 −5		wdl/structs.wdl
+12 −0		wdl/workflows/deepvariant/README.md
+33 −13		wdl/workflows/deepvariant/deepvariant.wdl
+1 −1		wdl/workflows/deepvariant/inputs.json