Adding AWS HealthOmics as a Module in "Play" tools

backend/dataall/base/cdkproxy/requirements.txt

@@ -1,7 +1,7 @@
 aws-cdk-lib==2.99.0
-boto3==1.24.85
-boto3-stubs==1.24.85
-botocore==1.27.85
+boto3==1.28.23
+boto3-stubs==1.28.23
+botocore==1.31.23
 cdk-nag==2.7.2
 constructs==10.0.73
 starlette==0.36.3

backend/dataall/core/environment/cdk/env_role_core_policies/service_policy.py

@@ -89,6 +89,7 @@ def generate_policies(self) -> [aws_iam.ManagedPolicy]:
                             'StringEquals': {
                                 'iam:PassedToService': [
                                     'glue.amazonaws.com',
+                                    'omics.amazonaws.com',
                                     'lambda.amazonaws.com',
                                     'sagemaker.amazonaws.com',
                                     'states.amazonaws.com',

backend/dataall/core/environment/cdk/environment_stack.py

@@ -487,6 +487,7 @@ def create_group_environment_role(self, group: EnvironmentGroup, id: str):
                 iam.ServicePrincipal('databrew.amazonaws.com'),
                 iam.ServicePrincipal('codebuild.amazonaws.com'),
                 iam.ServicePrincipal('codepipeline.amazonaws.com'),
+                iam.ServicePrincipal('omics.amazonaws.com'),
                 self.pivot_role,
             ),
         )

backend/dataall/modules/omics/__init__.py

@@ -0,0 +1,41 @@
+"""Contains the code related to HealthOmics"""
+
+import logging
+from typing import Set, List, Type
+
+from dataall.base.loader import ImportMode, ModuleInterface
+from dataall.modules.omics.db.omics_repository import OmicsRepository
+
+log = logging.getLogger(__name__)
+
+
+class OmicsApiModuleInterface(ModuleInterface):
+    """Implements ModuleInterface for omics GraphQl lambda"""
+
+    @staticmethod
+    def is_supported(modes: Set[ImportMode]) -> bool:
+        return ImportMode.API in modes
+
+    @staticmethod
+    def depends_on() -> List[Type['ModuleInterface']]:
+        from dataall.modules.s3_datasets import DatasetApiModuleInterface
+
+        return [DatasetApiModuleInterface]
+
+    def __init__(self):
+        import dataall.modules.omics.api
+
+        log.info('API of omics has been imported')
+
+
+class OmicsCdkModuleInterface(ModuleInterface):
+    """Implements ModuleInterface for omics ecs tasks"""
+
+    @staticmethod
+    def is_supported(modes: Set[ImportMode]) -> bool:
+        return ImportMode.CDK in modes
+
+    def __init__(self):
+        import dataall.modules.omics.cdk
+
+        log.info('API of Omics has been imported')

backend/dataall/modules/omics/api/__init__.py

@@ -0,0 +1,5 @@
+"""The package defines the schema for Omics Pipelines"""
+
+from dataall.modules.omics.api import input_types, mutations, queries, types, resolvers
+
+__all__ = ['types', 'input_types', 'queries', 'mutations', 'resolvers']

backend/dataall/modules/omics/api/enums.py

@@ -0,0 +1,6 @@
+from dataall.base.api.constants import GraphQLEnumMapper
+
+
+class OmicsWorkflowType(GraphQLEnumMapper):
+    PRIVATE = 'PRIVATE'
+    READY2RUN = 'READY2RUN'

backend/dataall/modules/omics/api/input_types.py

@@ -0,0 +1,32 @@
+"""The module defines GraphQL input types for Omics Runs"""
+
+from dataall.base.api import gql
+
+NewOmicsRunInput = gql.InputType(
+    name='NewOmicsRunInput',
+    arguments=[
+        gql.Field('environmentUri', type=gql.NonNullableType(gql.String)),
+        gql.Field('workflowUri', type=gql.NonNullableType(gql.String)),
+        gql.Field('label', type=gql.NonNullableType(gql.String)),
+        gql.Field('destination', type=gql.String),
+        gql.Field('parameterTemplate', type=gql.String),
+        gql.Field('SamlAdminGroupName', type=gql.NonNullableType(gql.String)),
+    ],
+)
+
+OmicsFilter = gql.InputType(
+    name='OmicsFilter',
+    arguments=[
+        gql.Argument(name='term', type=gql.String),
+        gql.Argument(name='page', type=gql.Integer),
+        gql.Argument(name='pageSize', type=gql.Integer),
+    ],
+)
+
+OmicsDeleteInput = gql.InputType(
+    name='OmicsDeleteInput',
+    arguments=[
+        gql.Argument(name='runUris', type=gql.NonNullableType(gql.ArrayType(gql.String))),
+        gql.Argument(name='deleteFromAWS', type=gql.Boolean),
+    ],
+)

backend/dataall/modules/omics/api/mutations.py

@@ -0,0 +1,20 @@
+"""The module defines GraphQL mutations for Omics Pipelines"""
+
+from dataall.base.api import gql
+from .resolvers import create_omics_run, delete_omics_run
+from .types import OmicsRun
+from .input_types import NewOmicsRunInput, OmicsDeleteInput
+
+createOmicsRun = gql.MutationField(
+    name='createOmicsRun',
+    type=OmicsRun,
+    args=[gql.Argument(name='input', type=gql.NonNullableType(NewOmicsRunInput))],
+    resolver=create_omics_run,
+)
+
+deleteOmicsRun = gql.MutationField(
+    name='deleteOmicsRun',
+    type=gql.Boolean,
+    args=[gql.Argument(name='input', type=gql.NonNullableType(OmicsDeleteInput))],
+    resolver=delete_omics_run,
+)

backend/dataall/modules/omics/api/queries.py

@@ -0,0 +1,27 @@
+"""The module defines GraphQL queries for Omics runs"""
+
+from dataall.base.api import gql
+from .resolvers import list_omics_runs, get_omics_workflow, list_omics_workflows
+from .types import OmicsRunSearchResults, OmicsWorkflow, OmicsWorkflows
+from .input_types import OmicsFilter
+
+listOmicsRuns = gql.QueryField(
+    name='listOmicsRuns',
+    args=[gql.Argument(name='filter', type=OmicsFilter)],
+    resolver=list_omics_runs,
+    type=OmicsRunSearchResults,
+)
+
+getOmicsWorkflow = gql.QueryField(
+    name='getOmicsWorkflow',
+    args=[gql.Argument(name='workflowUri', type=gql.NonNullableType(gql.String))],
+    type=OmicsWorkflow,
+    resolver=get_omics_workflow,
+)
+
+listOmicsWorkflows = gql.QueryField(
+    name='listOmicsWorkflows',
+    args=[gql.Argument(name='filter', type=OmicsFilter)],
+    type=OmicsWorkflows,
+    resolver=list_omics_workflows,
+)

backend/dataall/modules/omics/api/resolvers.py

@@ -0,0 +1,76 @@
+import logging
+from dataall.base.api.context import Context
+from dataall.base.db import exceptions
+from dataall.modules.omics.services.omics_service import OmicsService
+from dataall.modules.omics.db.omics_models import OmicsRun
+
+log = logging.getLogger(__name__)
+
+
+class RequestValidator:
+    """Aggregates all validation logic for operating with omics"""
+
+    @staticmethod
+    def required_uri(uri):
+        if not uri:
+            raise exceptions.RequiredParameter('URI')
+
+    @staticmethod
+    def validate_creation_request(data):
+        required = RequestValidator._required
+        if not data:
+            raise exceptions.RequiredParameter('data')
+        if not data.get('label'):
+            raise exceptions.RequiredParameter('name')
+
+        required(data, 'environmentUri')
+        required(data, 'SamlAdminGroupName')
+        required(data, 'workflowUri')
+        required(data, 'parameterTemplate')
+        required(data, 'destination')
+
+    @staticmethod
+    def _required(data: dict, name: str):
+        if not data.get(name):
+            raise exceptions.RequiredParameter(name)
+
+
+def create_omics_run(context: Context, source, input=None):
+    RequestValidator.validate_creation_request(input)
+    return OmicsService.create_omics_run(
+        uri=input['environmentUri'], admin_group=input['SamlAdminGroupName'], data=input
+    )
+
+
+def list_omics_runs(context: Context, source, filter: dict = None):
+    if not filter:
+        filter = {}
+    return OmicsService.list_user_omics_runs(filter)
+
+
+def list_omics_workflows(context: Context, source, filter: dict = None):
+    if not filter:
+        filter = {}
+    return OmicsService.list_omics_workflows(filter)
+
+
+def get_omics_workflow(context: Context, source, workflowUri: str = None):
+    RequestValidator.required_uri(workflowUri)
+    return OmicsService.get_omics_workflow(workflowUri)
+
+
+def delete_omics_run(context: Context, source, input):
+    RequestValidator.required_uri(input.get('runUris'))
+    return OmicsService.delete_omics_runs(uris=input.get('runUris'), delete_from_aws=input.get('deleteFromAWS', True))
+
+
+def resolve_omics_workflow(context, source: OmicsRun, **kwargs):
+    if not source:
+        return None
+    return OmicsService.get_omics_workflow(source.workflowUri)
+
+
+def resolve_omics_run_details(context, source: OmicsRun, **kwargs):
+    if not source:
+        return None
+    return OmicsService.get_omics_run_details_from_aws(source.runUri)

backend/dataall/modules/omics/api/types.py

@@ -0,0 +1,91 @@
+from dataall.base.api import gql
+from .resolvers import resolve_omics_workflow, resolve_omics_run_details
+from dataall.core.organizations.api.resolvers import resolve_organization_by_env
+from dataall.core.environment.api.resolvers import resolve_environment
+
+OmicsWorkflow = gql.ObjectType(
+    name='OmicsWorkflow',
+    fields=[
+        gql.Field(name='workflowUri', type=gql.String),
+        gql.Field(name='id', type=gql.String),
+        gql.Field(name='arn', type=gql.String),
+        gql.Field(name='name', type=gql.String),
+        gql.Field(name='label', type=gql.String),
+        gql.Field(name='type', type=gql.String),
+        gql.Field(name='description', type=gql.String),
+        gql.Field(name='parameterTemplate', type=gql.String),
+        gql.Field(name='environmentUri', type=gql.String),
+    ],
+)
+
+OmicsWorkflows = gql.ObjectType(
+    name='OmicsWorkflows',
+    fields=[
+        gql.Field(name='count', type=gql.Integer),
+        gql.Field(name='page', type=gql.Integer),
+        gql.Field(name='pages', type=gql.Integer),
+        gql.Field(name='hasNext', type=gql.Boolean),
+        gql.Field(name='hasPrevious', type=gql.Boolean),
+        gql.Field(name='nodes', type=gql.ArrayType(OmicsWorkflow)),
+    ],
+)
+
+OmicsRunStatus = gql.ObjectType(
+    name='OmicsRunStatus',
+    fields=[gql.Field(name='status', type=gql.String), gql.Field(name='statusMessage', type=gql.String)],
+)
+
+
+OmicsRun = gql.ObjectType(
+    name='OmicsRun',
+    fields=[
+        gql.Field('runUri', type=gql.ID),
+        gql.Field('environmentUri', type=gql.String),
+        gql.Field('organizationUri', type=gql.String),
+        gql.Field('name', type=gql.String),
+        gql.Field('label', type=gql.String),
+        gql.Field('description', type=gql.String),
+        gql.Field('tags', type=gql.ArrayType(gql.String)),
+        gql.Field('created', type=gql.String),
+        gql.Field('updated', type=gql.String),
+        gql.Field('owner', type=gql.String),
+        gql.Field('workflowUri', type=gql.String),
+        gql.Field('SamlAdminGroupName', type=gql.String),
+        gql.Field('parameterTemplate', type=gql.String),
+        gql.Field('outputDatasetUri', type=gql.String),
+        gql.Field('outputUri', type=gql.String),
+        gql.Field(
+            name='environment',
+            type=gql.Ref('Environment'),
+            resolver=resolve_environment,
+        ),
+        gql.Field(
+            name='organization',
+            type=gql.Ref('Organization'),
+            resolver=resolve_organization_by_env,
+        ),
+        gql.Field(
+            name='workflow',
+            type=OmicsWorkflow,
+            resolver=resolve_omics_workflow,
+        ),
+        gql.Field(
+            name='status',
+            type=OmicsRunStatus,
+            resolver=resolve_omics_run_details,
+        ),
+    ],
+)
+
+
+OmicsRunSearchResults = gql.ObjectType(
+    name='OmicsRunSearchResults',
+    fields=[
+        gql.Field(name='count', type=gql.Integer),
+        gql.Field(name='page', type=gql.Integer),
+        gql.Field(name='pages', type=gql.Integer),
+        gql.Field(name='hasNext', type=gql.Boolean),
+        gql.Field(name='hasPrevious', type=gql.Boolean),
+        gql.Field(name='nodes', type=gql.ArrayType(OmicsRun)),
+    ],
+)

backend/dataall/modules/omics/aws/__init__.py

@@ -0,0 +1 @@
+"""Contains code that send requests to AWS using SDK (boto3)"""