Create dataproc cluster task (#933)

requirements.in

@@ -1,7 +1,6 @@
-elasticsearch==7.9.1
-google-api-python-client>=1.8.0
 hail==0.2.132
-luigi>=3.4.0
+luigi==3.5.2
 gnomad==0.6.4
 aiofiles==24.1.0
 pydantic==2.8.2
+google-cloud-dataproc==5.14.0

requirements.txt

@@ -2,7 +2,7 @@
 # This file is autogenerated by pip-compile with Python 3.10
 # by the following command:
 #
-#    pip-compile requirements.in
+#    pip-compile --resolver=backtracking requirements.in
 #
 aiodns==2.0.0
     # via hail
@@ -28,25 +28,25 @@ avro==1.11.3
     # via hail
 azure-common==1.1.28
     # via azure-mgmt-storage
-azure-core==1.31.0
+azure-core==1.32.0
     # via
     #   azure-identity
     #   azure-mgmt-core
     #   azure-storage-blob
     #   msrest
 azure-identity==1.19.0
     # via hail
-azure-mgmt-core==1.4.0
+azure-mgmt-core==1.5.0
     # via azure-mgmt-storage
 azure-mgmt-storage==20.1.0
     # via hail
 azure-storage-blob==12.23.1
     # via hail
 bokeh==3.3.4
     # via hail
-boto3==1.35.48
+boto3==1.35.53
     # via hail
-botocore==1.35.48
+botocore==1.35.53
     # via
     #   boto3
     #   hail
@@ -55,7 +55,6 @@ cachetools==5.5.0
     # via google-auth
 certifi==2024.8.30
     # via
-    #   elasticsearch
     #   msrest
     #   requests
 cffi==1.17.1
@@ -86,10 +85,6 @@ deprecated==1.2.14
     # via hail
 dill==0.3.9
     # via hail
-docutils==0.21.2
-    # via python-daemon
-elasticsearch==7.9.1
-    # via -r requirements.in
 exceptiongroup==1.2.2
     # via ipython
 executing==2.1.0
@@ -101,38 +96,44 @@ frozenlist==1.5.0
     #   hail
 gnomad==0.6.4
     # via -r requirements.in
-google-api-core==2.21.0
-    # via google-api-python-client
-google-api-python-client==2.149.0
-    # via -r requirements.in
+google-api-core[grpc]==2.22.0
+    # via google-cloud-dataproc
 google-auth==2.35.0
     # via
     #   google-api-core
-    #   google-api-python-client
-    #   google-auth-httplib2
     #   google-auth-oauthlib
+    #   google-cloud-dataproc
     #   hail
-google-auth-httplib2==0.2.0
-    # via google-api-python-client
 google-auth-oauthlib==0.8.0
     # via hail
-googleapis-common-protos==1.65.0
+google-cloud-dataproc==5.14.0
+    # via -r requirements.in
+googleapis-common-protos[grpc]==1.65.0
+    # via
+    #   google-api-core
+    #   grpc-google-iam-v1
+    #   grpcio-status
+grpc-google-iam-v1==0.13.1
+    # via google-cloud-dataproc
+grpcio==1.67.1
+    # via
+    #   google-api-core
+    #   googleapis-common-protos
+    #   grpc-google-iam-v1
+    #   grpcio-status
+grpcio-status==1.48.2
     # via google-api-core
 hail==0.2.132
     # via -r requirements.in
 hdbscan==0.8.39
     # via gnomad
-httplib2==0.22.0
-    # via
-    #   google-api-python-client
-    #   google-auth-httplib2
 humanize==1.1.0
     # via hail
 idna==3.10
     # via
     #   requests
     #   yarl
-ipython==8.28.0
+ipython==8.29.0
     # via ipywidgets
 ipywidgets==8.1.5
     # via gnomad
@@ -218,11 +219,16 @@ prompt-toolkit==3.0.48
 propcache==0.2.0
     # via yarl
 proto-plus==1.25.0
-    # via google-api-core
+    # via
+    #   google-api-core
+    #   google-cloud-dataproc
 protobuf==3.20.2
     # via
     #   google-api-core
+    #   google-cloud-dataproc
     #   googleapis-common-protos
+    #   grpc-google-iam-v1
+    #   grpcio-status
     #   hail
     #   proto-plus
 ptyprocess==0.7.0
@@ -251,11 +257,9 @@ pygments==2.18.0
     #   rich
 pyjwt[crypto]==2.9.0
     # via msal
-pyparsing==3.2.0
-    # via httplib2
 pyspark==3.5.3
     # via hail
-python-daemon==3.0.1
+python-daemon==3.1.0
     # via luigi
 python-dateutil==2.9.0.post0
     # via
@@ -348,12 +352,9 @@ typing-extensions==4.12.2
     #   typer
 tzdata==2024.2
     # via pandas
-uritemplate==4.1.1
-    # via google-api-python-client
 urllib3==2.2.3
     # via
     #   botocore
-    #   elasticsearch
     #   requests
 uvloop==0.21.0
     # via hail
@@ -365,7 +366,7 @@ wrapt==1.16.0
     # via deprecated
 xyzservices==2024.9.0
     # via bokeh
-yarl==1.16.0
+yarl==1.17.1
     # via aiohttp
 
 # The following packages are considered to be unsafe in a requirements file:

v03_pipeline/bin/dataproc_vep_init.bash

@@ -14,7 +14,7 @@
 set -x
 
 export PROJECT="$(gcloud config get-value project)"
-export ENVIRONMENT="$(/usr/share/google/get_metadata_value attributes/ENVIRONMENT)"
+export DEPLOYMENT_TYPE="$(/usr/share/google/get_metadata_value attributes/DEPLOYMENT_TYPE)"
 export REFERENCE_GENOME="$(/usr/share/google/get_metadata_value attributes/REFERENCE_GENOME)"
 export PIPELINE_RUNNER_APP_VERSION="$(/usr/share/google/get_metadata_value attributes/PIPELINE_RUNNER_APP_VERSION)"
 
@@ -53,10 +53,10 @@ EOF
 gcc -Wall -Werror -O2 /vep.c -o /vep
 chmod u+s /vep
 
-gcloud storage cp gs://seqr-pipeline-runner-builds/$ENVIRONMENT/$PIPELINE_RUNNER_APP_VERSION/bin/download_vep_reference_data.bash /download_vep_reference_data.bash
+gcloud storage cp gs://seqr-pipeline-runner-builds/$DEPLOYMENT_TYPE/$PIPELINE_RUNNER_APP_VERSION/bin/download_vep_reference_data.bash /download_vep_reference_data.bash
 chmod +x /download_vep_reference_data.bash
 ./download_vep_reference_data.bash $REFERENCE_GENOME
 
-gcloud storage cp gs://seqr-pipeline-runner-builds/$ENVIRONMENT/$PIPELINE_RUNNER_APP_VERSION/bin/vep /vep.bash
+gcloud storage cp gs://seqr-pipeline-runner-builds/$DEPLOYMENT_TYPE/$PIPELINE_RUNNER_APP_VERSION/bin/vep /vep.bash
 chmod +x /vep.bash
 

v03_pipeline/lib/model/dataset_type.py

@@ -361,6 +361,10 @@ def lookup_table_annotation_fns(self) -> list[Callable[..., hl.Expression]]:
     def should_send_to_allele_registry(self):
         return self == DatasetType.SNV_INDEL
 
+    @property
+    def requires_dataproc(self):
+        return self == DatasetType.SNV_INDEL
+
     @property
     def should_export_to_vcf(self):
         return self == DatasetType.SV

v03_pipeline/lib/model/environment.py

@@ -1,5 +1,6 @@
 import os
 from dataclasses import dataclass
+from typing import Literal
 
 # NB: using os.environ.get inside the dataclass defaults gives a lint error.
 GRCH37_TO_GRCH38_LIFTOVER_REF_PATH = os.environ.get(
@@ -41,6 +42,14 @@
     'CLINGEN_ALLELE_REGISTRY_PASSWORD',
     '',
 )
+DEPLOYMENT_TYPE = os.environ.get('DEPLOYMENT_TYPE', 'prod')
+GCLOUD_DATAPROC_SECONDARY_WORKERS = int(
+    os.environ.get('GCLOUD_DATAPROC_SECONDARY_WORKERS', '5'),
+)
+GCLOUD_PROJECT = os.environ.get('GCLOUD_PROJECT')
+GCLOUD_ZONE = os.environ.get('GCLOUD_ZONE')
+GCLOUD_REGION = os.environ.get('GCLOUD_REGION')
+PIPELINE_RUNNER_APP_VERSION = os.environ.get('PIPELINE_RUNNER_APP_VERSION', 'latest')
 
 # Feature Flags
 ACCESS_PRIVATE_REFERENCE_DATASETS = (
@@ -62,7 +71,12 @@ class Env:
     CHECK_SEX_AND_RELATEDNESS: bool = CHECK_SEX_AND_RELATEDNESS
     CLINGEN_ALLELE_REGISTRY_LOGIN: str | None = CLINGEN_ALLELE_REGISTRY_LOGIN
     CLINGEN_ALLELE_REGISTRY_PASSWORD: str | None = CLINGEN_ALLELE_REGISTRY_PASSWORD
+    DEPLOYMENT_TYPE: Literal['dev', 'prod'] = DEPLOYMENT_TYPE
     EXPECT_WES_FILTERS: bool = EXPECT_WES_FILTERS
+    GCLOUD_DATAPROC_SECONDARY_WORKERS: str = GCLOUD_DATAPROC_SECONDARY_WORKERS
+    GCLOUD_PROJECT: str | None = GCLOUD_PROJECT
+    GCLOUD_ZONE: str | None = GCLOUD_ZONE
+    GCLOUD_REGION: str | None = GCLOUD_REGION
     GRCH37_TO_GRCH38_LIFTOVER_REF_PATH: str = GRCH37_TO_GRCH38_LIFTOVER_REF_PATH
     GRCH38_TO_GRCH37_LIFTOVER_REF_PATH: str = GRCH38_TO_GRCH37_LIFTOVER_REF_PATH
     HAIL_BACKEND_SERVICE_HOSTNAME: str | None = HAIL_BACKEND_SERVICE_HOSTNAME
@@ -71,6 +85,7 @@ class Env:
     HAIL_SEARCH_DATA_DIR: str = HAIL_SEARCH_DATA_DIR
     INCLUDE_PIPELINE_VERSION_IN_PREFIX: bool = INCLUDE_PIPELINE_VERSION_IN_PREFIX
     LOADING_DATASETS_DIR: str = LOADING_DATASETS_DIR
+    PIPELINE_RUNNER_APP_VERSION: str = PIPELINE_RUNNER_APP_VERSION
     PRIVATE_REFERENCE_DATASETS_DIR: str = PRIVATE_REFERENCE_DATASETS_DIR
     REFERENCE_DATASETS_DIR: str = REFERENCE_DATASETS_DIR
     SHOULD_TRIGGER_HAIL_BACKEND_RELOAD: bool = SHOULD_TRIGGER_HAIL_BACKEND_RELOAD