feat: adding cadd-rest-api support via docker-compose.override.yml (#48)

Author: holtgrewe

.gitignore

@@ -6,6 +6,7 @@
 ## Docker Compose
 # Configuration file.
 .env
+/*.env
 .secrets/
 
 # The Docker Compose "override" YAML file.

docker-compose.override.yml-cadd

@@ -0,0 +1,56 @@
+# Docker Compose Override YAML fragment that provides a CADD REST API
+# # Server
+#
+# Quick Manual:
+#
+# - merge with docker-compose.override.yml
+# - place data in the right place
+
+# == Re-useable Definitions ==================================================
+
+# Default service definition for all (incl. postgres/redis/...)
+x-service-default: &service_default
+  networks:
+    - varfish
+  restart: unless-stopped
+
+# == Overide Definitions =====================================================
+
+services:
+  # -- CADD REST API ---------------------------------------------------------
+  # 
+  # REST API server for CADD sequence variant annotation.
+
+  cadd-rest-api-server:
+    <<: *service_default
+    image: "${image_cadd_name-bihealth/cadd-rest-api}:${image_cadd_version-0.3.4-0}"
+    env_file: cadd-rest-api.env
+    command:
+      - wsgi
+    volumes:
+      - type: bind
+        source: ${volumes_basedir:-./.dev/volumes}/cadd-rest-api/db
+        target: /data/db
+        read_only: false
+
+  cadd-rest-api-worker:
+    <<: *service_default
+    deploy:
+      replicas: 5
+    image: "${image_cadd_name-bihealth/cadd-rest-api}:${image_cadd_version-0.3.4-0}"
+    env_file: cadd-rest-api.env
+    command:
+      - celeryd
+    volumes:
+      - type: bind
+        source: ${volumes_basedir:-./.dev/volumes}/cadd-rest-api/data/annotations
+        target: /opt/miniconda3/share/cadd-scripts-1.6-0/data/annotations
+        read_only: true
+      - type: bind
+        source: ${volumes_basedir:-./.dev/volumes}/cadd-rest-api/data/prescored
+        target: /opt/miniconda3/share/cadd-scripts-1.6-0/data/prescored
+        read_only: true
+      - type: bind
+        source: ${volumes_basedir:-./.dev/volumes}/cadd-rest-api/db
+        target: /data/db
+        read_only: false

docker-compose.yml

@@ -105,6 +105,7 @@ services:
       - "traefik.http.routers.varfish-web.middlewares=xforward"
       - "traefik.http.routers.varfish-web.rule=HostRegexp(`{catchall:.+}`)"
       - "traefik.http.services.varfish-web.loadbalancer.server.port=8080"
+      - "traefik.http.routers.varfish-web.tls=true"
 
   varfish-celerybeat:
     <<: *service_varfish_server_default

download-data.sh

@@ -242,7 +242,7 @@ EOF
 mkdir -p $DATA_DIR/download
 # Download each entry from download list.  Note that we support commenting
 # out lines with a leading "#".
-grep -v ^# /tmp/download-list.txt >/tmp/download-list.nocomment.txt
+grep -v ^# /tmp/download-list.txt | grep -v grch37 >/tmp/download-list.nocomment.txt
 while read -r line; do
     # Create the download directory.
     run mkdir -p $DATA_DIR/download/$line
@@ -251,6 +251,7 @@ while read -r line; do
     run s5cmd \
         --endpoint-url=$S3_ENDPOINT_URL \
         --no-sign-request \
+	--no-verify-ssl \
         sync \
         "s3://varfish-public/$(prefix_for $line)/$line/*" \
         $DATA_DIR/download/$line \

utils/cadd-rest-api/cadd-rest-api.env

@@ -0,0 +1,20 @@
+# Environment for cadd-rest-api related containers.
+
+LC_ALL=en_US.UTF-8
+
+DATABASE_URL=/data/db/cadd_rest_api.db
+
+# We can share one redis instance but not the database with another celery.
+CELERY_BROKER_URL=redis://redis:6379/1
+# Only one CADD process can be run per CADD container because of using
+# Snakemake.
+CELERY_WORKERS=1
+
+DJANGO_ALLOWED_HOSTS="*"
+DJANGO_SECRET_KEY="__DJANGO_SECRET_KEY__"
+DJANGO_SETTINGS_MODULE="config.settings.production"
+DJANGO_SECURE_SSL_REDIRECT=0
+
+CADD_SH=/opt/miniconda3/share/cadd-scripts-1.6-0/CADD.sh
+CADD_CONDA=/opt/miniconda3/bin/activate
+CADD_TIMEOUT=300