Add taxonomy in the extraction log file name

EBIvariation · Jan 5, 2024 · c904ad5 · c904ad5
1 parent 2aeb9dc
commit c904ad5
Show file tree

Hide file tree

Showing 2 changed files with 17 additions and 11 deletions.
diff --git a/eva_assembly_ingestion/nextflow/remap_cluster.nf b/eva_assembly_ingestion/nextflow/remap_cluster.nf
@@ -137,7 +137,7 @@ process extract_vcf_from_mongo {
     output:
     // Store both vcfs (eva and dbsnp), emit: one channel
     path '*.vcf', emit: source_vcfs
-    path "${params.source_assembly_accession}_vcf_extractor.log", emit: log_filename
+    path "${params.source_assembly_accession}_${taxonomy}_vcf_extractor.log", emit: log_filename
 
     publishDir "$params.output_dir/logs", overwrite: true, mode: "copy", pattern: "*.log*"
 
@@ -147,7 +147,7 @@ process extract_vcf_from_mongo {
         --parameters.fasta=${source_fasta} \
         --parameters.assemblyReportUrl=file:${source_report} \
         --parameters.taxonomy=${taxonomy}
-        > ${params.source_assembly_accession}_vcf_extractor.log
+        > ${params.source_assembly_accession}_${taxonomy}_vcf_extractor.log
     """
 }
 

diff --git a/tests/nextflow-tests/run_tests.sh b/tests/nextflow-tests/run_tests.sh
@@ -26,15 +26,21 @@ nextflow run ${SOURCE_DIR}/eva_assembly_ingestion/nextflow/remap_cluster.nf -par
 	 --remapping_required 1 \
 	 --memory 2
 
-ls ${SCRIPT_DIR}/output/dbsnp/GCA_0000001_dbsnp_remapped.vcf \
-   ${SCRIPT_DIR}/output/dbsnp/GCA_0000001_dbsnp_remapped_unmapped.vcf \
-   ${SCRIPT_DIR}/output/dbsnp/GCA_0000001_dbsnp_remapped_counts.yml \
-   ${SCRIPT_DIR}/output/eva/GCA_0000001_eva_remapped.vcf \
-   ${SCRIPT_DIR}/output/eva/GCA_0000001_eva_remapped_unmapped.vcf \
-   ${SCRIPT_DIR}/output/eva/GCA_0000001_eva_remapped_counts.yml
-
-# Test we have 7 log files in the logs directory (1 extraction, 2 ingestion, 3 clustering, 1 backpropagate)
-[[ $(find ${SCRIPT_DIR}/output/logs/ -type f -name "*.log" | wc -l) -eq 7 ]]
+ls ${SCRIPT_DIR}/output/dbsnp/GCA_0000001_1233_dbsnp_remapped.vcf \
+   ${SCRIPT_DIR}/output/dbsnp/GCA_0000001_1233_dbsnp_remapped_unmapped.vcf \
+   ${SCRIPT_DIR}/output/dbsnp/GCA_0000001_1233_dbsnp_remapped_counts.yml \
+   ${SCRIPT_DIR}/output/eva/GCA_0000001_1233_eva_remapped.vcf \
+   ${SCRIPT_DIR}/output/eva/GCA_0000001_1233_eva_remapped_unmapped.vcf \
+   ${SCRIPT_DIR}/output/eva/GCA_0000001_1233_eva_remapped_counts.yml \
+   ${SCRIPT_DIR}/output/dbsnp/GCA_0000001_1234_dbsnp_remapped.vcf \
+   ${SCRIPT_DIR}/output/dbsnp/GCA_0000001_1234_dbsnp_remapped_unmapped.vcf \
+   ${SCRIPT_DIR}/output/dbsnp/GCA_0000001_1234_dbsnp_remapped_counts.yml \
+   ${SCRIPT_DIR}/output/eva/GCA_0000001_1234_eva_remapped.vcf \
+   ${SCRIPT_DIR}/output/eva/GCA_0000001_1234_eva_remapped_unmapped.vcf \
+   ${SCRIPT_DIR}/output/eva/GCA_0000001_1234_eva_remapped_counts.yml
+
+# Test we have 10 log files in the logs directory (2 extraction, 4 ingestion, 3 clustering, 1 backpropagate)
+[[ $(find ${SCRIPT_DIR}/output/logs/ -type f -name "*.log" | wc -l) -eq 10 ]]
 
 # Test we have 1 rs_report in the logs directory
 [[ $(find ${SCRIPT_DIR}/output/logs/ -type f -name "*.txt" | wc -l) -eq 1 ]]