feat(IPVC-2386): create second docker-compose yaml for one off tasks,…

… move workflow and script to misc directory, add misc directory to dockerfile

Dockerfile

@@ -17,6 +17,7 @@ RUN pip install pysam
 WORKDIR /opt/repos/uta/
 COPY pyproject.toml ./
 COPY etc ./etc
+COPY misc ./misc
 COPY sbin ./sbin
 COPY src ./src
 RUN pip install -e .[dev]

docker-compose.yml

@@ -20,15 +20,6 @@ services:
       - ${UTA_ETL_LOG_DIR}:/uta-extract/logs
     working_dir: /opt/repos/uta
     network_mode: host
-  uta-extract-historical:
-    image: uta-update
-    command: sbin/uta-extract-historical /ncbi-dir /uta-extract/work /uta-extract/logs
-    volumes:
-      - ${UTA_ETL_NCBI_DIR}:/ncbi-dir
-      - ${UTA_ETL_WORK_DIR}:/uta-extract/work
-      - ${UTA_ETL_LOG_DIR}:/uta-extract/logs
-    working_dir: /opt/repos/uta
-    network_mode: host
   seqrepo-load:
     image: uta-update
     command: sbin/seqrepo-load /usr/local/share/seqrepo ${UTA_ETL_OLD_SEQREPO_VERSION} /seqrepo-load/work /seqrepo-load/logs

misc/refseq-historical-backfill/docker-compose-backfill.yml

@@ -0,0 +1,14 @@
+# docker compose file for the RefSeq historical backfill procedure
+
+version: '3'
+
+services:
+  uta-extract-historical:
+    image: uta-update
+    command: misc/refseq-historical-backfill/uta-extract-historical /ncbi-dir /uta-extract/work /uta-extract/logs
+    volumes:
+      - ${UTA_ETL_NCBI_DIR}:/ncbi-dir
+      - ${UTA_ETL_WORK_DIR}:/uta-extract/work
+      - ${UTA_ETL_LOG_DIR}:/uta-extract/logs
+    working_dir: /opt/repos/uta
+    network_mode: host

sbin/uta-extract-historical → misc/refseq-historical-backfill/uta-extract-historical

@@ -38,7 +38,8 @@ file_path="$relative_path/GCF_000001405.40-RS_2023_03_knownrefseq_alns.gff.gz"
 download_ncbi_file $file_path $ncbi_dir
 
 # extract intermediate files from genbank file
-python sbin/ncbi_extract_gbff.py "$ncbi_dir/$relative_path/GCF_000001405.40-RS_2023_03_knownrefseq_rna.gbff.gz" \
+python misc/refseq-historical-backfill/ncbi_extract_gbff.py \
+  "$ncbi_dir/$relative_path/GCF_000001405.40-RS_2023_03_knownrefseq_rna.gbff.gz" \
   --output_dir "$working_dir" 2>&1 | tee "$log_dir/ncbi-parse-historical-ggbb.log"
 
 # extract exonset intermediate file from gff file