Merge branch 'master' into 'scp-sample'

# Conflicts:
#   prepare_vm.sh

prepare_vm.sh

@@ -4,7 +4,7 @@ echo "#### Updating Ubuntu dist..."
 sudo apt update && sudo apt -y upgrade
 
 echo "#### Installing some Ubuntu packages..."
-sudo apt -y install unzip build-essential libncurses5-dev zlib1g-dev libbz2-dev liblzma-dev p7zip-full
+sudo apt -y install unzip build-essential libncurses5-dev zlib1g-dev libbz2-dev liblzma-dev p7zip-full pigz
 sudo apt -y install neovim
 
 echo "#### Installing samtools 1.10..."

wgbs-pipeline.sh

@@ -2,7 +2,7 @@
 
 # Some variables
 MT_chr_name="chrM"
-LAMBDA_chr_name="chrL"
+LAMBDA_chr_name="NC_001416.1"
 
 # Get pipeline folder
 PIPELINEPATH="$( cd "$(dirname "$0")" > /dev/null 2>&1 ; pwd -P )"
@@ -43,7 +43,23 @@ SUBSET_READS=false
 
 ########################################
 
-
+# Check that both MT chromosome and lambda are present in the reference
+if ! grep -m1 -q "^>${MT_chr_name}" ${b38_ref}/Bisulfite_Genome/CT_conversion/genome_mfa.CT_conversion.fa; then
+    echo "ERROR: \"${MT_chr_name}\" does not exist in ${b38_ref}"
+    exit 1
+fi
+if grep -m1 -q "^>${MT_chr_name}" ${nuc_ref}/Bisulfite_Genome/CT_conversion/genome_mfa.CT_conversion.fa; then
+    echo "ERROR: \"${MT_chr_name}\" is present in the nuclear-only reference: ${nuc_ref}"
+    exit 1
+fi
+if ! grep -m1 -q "^>${MT_chr_name}" ${mt_ref}/Bisulfite_Genome/CT_conversion/genome_mfa.CT_conversion.fa; then
+    echo "ERROR: \"${MT_chr_name}\" does not exist in ${mt_ref}"
+    exit 1
+fi
+if ! grep -m1 -q "^>${LAMBDA_chr_name}" ${b38_ref}/Bisulfite_Genome/CT_conversion/genome_mfa.CT_conversion.fa; then
+    echo "ERROR: \"${LAMBDA_chr_name}\" does not exist in ${b38_ref}"
+    exit 1
+fi
 
 
 # Exit when any command fails
@@ -291,23 +307,24 @@ else
     echo "DONE"
 fi
 
+
 echo -ne "     1.5 Extracting MT and Lambda methylation... "
 
-CMD="cat ${output_dir}/CpG_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${MT_chr_name} | gzip -nc > \
+CMD="cat ${output_dir}/CpG_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${MT_chr_name} | pigz -nc > \
     ${output_dir}/CpG_context_chrM_${sample_id}.dedup.txt.gz && \
-    cat ${output_dir}/CpG_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${LAMBDA_chr_name} | gzip -nc > \
+    cat ${output_dir}/CpG_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${LAMBDA_chr_name} | pigz -nc > \
     ${output_dir}/CpG_context_LAMBDA_${sample_id}.dedup.txt.gz && \
-    gzip ${output_dir}/CpG_context_b38_${sample_id}.dedup.txt \n\
-cat ${output_dir}/CHH_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${MT_chr_name} | gzip -nc > \
+    pigz ${output_dir}/CpG_context_b38_${sample_id}.dedup.txt \n\
+cat ${output_dir}/CHH_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${MT_chr_name} | pigz -nc > \
     ${output_dir}/CHH_context_chrM_${sample_id}.dedup.txt.gz && \
-    cat ${output_dir}/CHH_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${LAMBDA_chr_name} | gzip -nc > \
+    cat ${output_dir}/CHH_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${LAMBDA_chr_name} | pigz -nc > \
     ${output_dir}/CHH_context_LAMBDA_${sample_id}.dedup.txt.gz && \
-    gzip ${output_dir}/CHH_context_b38_${sample_id}.dedup.txt \n\
-cat ${output_dir}/CHG_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${MT_chr_name} | gzip -nc > \
+    pigz ${output_dir}/CHH_context_b38_${sample_id}.dedup.txt \n\
+cat ${output_dir}/CHG_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${MT_chr_name} | pigz -nc > \
     ${output_dir}/CHG_context_chrM_${sample_id}.dedup.txt.gz && \
-    cat ${output_dir}/CHG_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${LAMBDA_chr_name} | gzip -nc > \
+    cat ${output_dir}/CHG_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${LAMBDA_chr_name} | pigz -nc > \
     ${output_dir}/CHG_context_LAMBDA_${sample_id}.dedup.txt.gz && \
-    gzip ${output_dir}/CHG_context_b38_${sample_id}.dedup.txt \n\
+    pigz ${output_dir}/CHG_context_b38_${sample_id}.dedup.txt \n\
     zcat ${output_dir}/C??_context_LAMBDA_${sample_id}.dedup.txt.gz | python3 ${meth_prop} > \
     ${output_dir}/Lambda_meth_prop_${sample_id}.txt"
 echo -e ${CMD} > ${log_dir}/1_5_extract_MT.cmd