Compression testing

.gitignore

@@ -10,3 +10,4 @@ cloudgene.report.json
 *.jar
 .idea
 scratch
+imputationserver2.iml

modules/local/compression/compression_encryption_vcf.nf

@@ -24,27 +24,38 @@ process COMPRESSION_ENCRYPTION_VCF {
     def panel_version = params.refpanel.id
     def scale   = params.encryption.thread_scale ?: 1
     def threads = Math.max(1, Math.floor(task.cpus * scale) as int)
+    def first_vcf = getFirstFile(imputed_vcf_data)
+
 
     """
     # concat info files
     bcftools concat --threads ${threads} -n ${info_joined} -o ${info_name} -Oz
 
-    # concat dosage files and update header
-    bcftools concat --threads ${threads} -n ${imputed_joined} -o intermediate_${imputed_name} -Oz
-
     # annotate files
     if [[ "${params.encryption.annotate}" = true ]]
     then
-        echo "##mis_pipeline=${workflow.manifest.version}" > add_header.txt
-        echo "##mis_phasing=${params.phasing.engine}" >> add_header.txt
-        echo "##mis_panel=${panel_version}" >> add_header.txt
-
-        bcftools annotate --threads ${threads} -h add_header.txt intermediate_${imputed_name} -o ${imputed_name} -Oz
+        #Extracting the original header to append the new one
+        bcftools view -h "${first_vcf}" > header_from_chunk.txt
+
+        # build final header: keep ## lines, add custom lines, then #CHROM last
+         awk -v pipeline="${workflow.manifest.version}" \
+              -v phasing="${params.phasing.engine}" \
+              -v panel="${panel_version}" '
+          /^#CHROM/ {
+              print "##mis_pipeline=" pipeline
+              print "##mis_phasing=" phasing
+              print "##mis_panel=" panel
+          }
+          { print }
+          ' header_from_chunk.txt > final_header.txt
+
+        # concat dosage files and update header
+        bcftools concat --threads ${threads} -n ${imputed_joined} -Oz | bcftools reheader -h final_header.txt -o ${imputed_name}
 
-        rm intermediate_${imputed_name}
     else
+        bcftools concat --threads ${threads} -n ${imputed_joined} -o intermediate_${imputed_name} -Oz
         mv intermediate_${imputed_name} ${imputed_name}
-    fi
+   fi
 
     # write meta files
     if [[ "${params.imputation.meta}" = true ]]
@@ -90,3 +101,8 @@ def compareFilenames(a, b) {
 def processFileList(fileList) {
     return fileList.sort { a, b -> compareFilenames(a, b) }.join(" ")
 }
+
+def getFirstFile(fileList) {
+    def sorted = fileList.sort { a, b -> compareFilenames(a, b) }
+    return sorted[0]
+}