Fixed number of cores for alignment based on RAM

wgbs-pipeline.sh

@@ -34,12 +34,12 @@ psrecord="psrecord"
 ########################################
 
 # To activate CPU and RAM monitoring
-MONITOR=false
-#MONITOR=true
+#MONITOR=false
+MONITOR=true
 
 # To subset reads to 10,000 (debug)
-SUBSET_READS=false
-#SUBSET_READS=true
+#SUBSET_READS=false
+SUBSET_READS=true
 
 ########################################
 
@@ -53,6 +53,8 @@ set -e
 
 # Get number of processors in the machine
 N_proc=$( grep -c processor /proc/cpuinfo )
+# Get RAM in the machine
+RAM=$( free -g | grep '^Mem' | awk '{print $2}' )
 
 # Ensure reference files/indices are there, download and index otherwise
 if [[ ! -s ${b38_ref}/Bisulfite_Genome/GA_conversion/genome_mfa.GA_conversion.fa ]]; then
@@ -159,8 +161,8 @@ if $MONITOR; then
 fi
 
 if $SUBSET_READS; then
-    SUBSET_READS="-u 10000"
-    echo "SUBSETTING TO 10000 READS ONLY!!!!"
+    SUBSET_READS="-u 100000"
+    echo "SUBSETTING TO 100000 READS ONLY!!!!"
     echo
 fi
 
@@ -176,8 +178,10 @@ echo "##################################################"
 
  
 echo -ne "     1.1 Aligning... "
-cpus=`expr ${N_proc} / 4`
-#CMD="${bismark} ${b38_ref} -1 ${R1} -2 ${R2} --multicore ${cpus} -u 100000 \
+
+cpus=`expr ${RAM} / 16`
+if [[ $cpus -eq 0 ]]; then cpus=1; fi
+
 CMD="${bismark} ${b38_ref} -1 ${R1} -2 ${R2} --multicore ${cpus} ${SUBSET_READS} \
     -o ${output_dir} --prefix b38 --path_to_bowtie ${bowtie_path} \
     --samtools_path ${samtools_path} --temp_dir ${tmp_dir} > \
@@ -199,6 +203,7 @@ else
 fi
 
 echo -ne "     1.2 Merging bamfiles... "
+
 bamfile_list=""
 IFS=',' read -r -a array <<< "${R1}"
 for element in "${array[@]}"; do
@@ -207,9 +212,7 @@ for element in "${array[@]}"; do
     bamfile_list="${bamfile_list} ${bamfile}"
 done
 
-cpus=`expr ${N_proc} / 2`
-#CMD="${samtools} merge -n -r -@ ${cpus} ${output_dir}/b38_${sample_id}.bam ${bamfile_list} > \
-CMD="${samtools} cat -@ ${cpus} -o ${output_dir}/b38_${sample_id}.bam ${bamfile_list} > \
+CMD="${samtools} cat -o ${output_dir}/b38_${sample_id}.bam ${bamfile_list} > \
     ${log_dir}/1_2_merge.stdout 2> \
     ${log_dir}/1_2_merge.stderr"
 echo -e ${CMD} > ${log_dir}/1_2_merge.cmd
@@ -234,6 +237,7 @@ fi
 
 
 echo -ne "     1.3 Dedupping alignment... "
+
 CMD="${deduplicate_bismark} --paired --samtools_path ${samtools_path} \
     --output_dir ${output_dir} ${output_dir}/b38_${sample_id}.bam > \
     ${log_dir}/1_3_dedup.stdout 2> \
@@ -257,10 +261,11 @@ if [ -s "${output_dir}/b38_${sample_id}.deduplicated.bam" ]; then
     mv ${output_dir}/b38_${sample_id}.deduplicated.bam ${output_dir}/b38_${sample_id}.dedup.bam
 fi
 
-exit 1
 
 echo -ne "     1.4 Extracting methylation... "
+
 cpus=`expr ${N_proc} / 3`
+
 CMD="${bismark_meth_extract} -p --comprehensive --output ${output_dir} \
     --multicore ${cpus} \
     --samtools_path ${samtools_path} \
@@ -283,6 +288,7 @@ else
 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 > \
     ${output_dir}/CpG_context_chrM_${sample_id}.dedup.txt.gz && \
     cat ${output_dir}/CpG_context_b38_${sample_id}.dedup.txt | python3 ${ext_chr_srt} ${LAMBDA_chr_name} | gzip -nc > \
@@ -317,6 +323,8 @@ fi
 
 echo
 
+exit 1
+
 
 
 
@@ -332,8 +340,10 @@ echo "##################################################"
 
  
 echo -ne "     2.1 Aligning... "
-cpus=`expr ${N_proc} / 4`
-#CMD="${bismark} ${mt_ref} -1 ${R1} -2 ${R2} --multicore ${cpus} -u 100000 \
+
+cpus=`expr ${RAM} / 16`
+if [[ $cpus -eq 0 ]]; then cpus=1; fi
+
 CMD="${bismark} ${mt_ref} -1 ${R1} -2 ${R2} --multicore ${cpus} ${SUBSET_READS} \
     -o ${output_dir} --prefix MT --path_to_bowtie ${bowtie_path} \
     --samtools_path ${samtools_path} --temp_dir ${tmp_dir} > \
@@ -355,6 +365,9 @@ else
 fi
 
 echo -ne "     2.2 Merging bamfiles... "
+
+cpus=`expr ${N_proc} / 2`
+
 bamfile_list=""
 IFS=',' read -r -a array <<< "${R1}"
 for element in "${array[@]}"; do
@@ -363,7 +376,6 @@ for element in "${array[@]}"; do
     bamfile_list="${bamfile_list} ${bamfile}"
 done
 
-cpus=`expr ${N_proc} / 2`
 CMD="${samtools} merge -n -r -@ ${cpus} ${output_dir}/MT_${sample_id}.bam ${bamfile_list} > \
     ${log_dir}/2_2_merge.stdout 2> \
     ${log_dir}/2_2_merge.stderr"
@@ -414,7 +426,9 @@ fi
 
 
 echo -ne "     2.4 Extracting methylation... "
+
 cpus=`expr ${N_proc} / 3`
+
 CMD="${bismark_meth_extract} -p --comprehensive --output ${output_dir} \
     --gzip --multicore ${cpus} \
     --samtools_path ${samtools_path} \
@@ -456,8 +470,11 @@ echo "##################################################"
  
  
 echo -ne "     3.1 Aligning vs nuclear DNA... "
-cpus=`expr ${N_proc} / 4`
-#CMD="${bismark} ${nuc_ref} -1 ${R1} -2 ${R2} --multicore ${cpus} -u 100000 \
+
+cpus=`expr ${RAM} / 16`
+if [[ $cpus -eq 0 ]]; then cpus=1; fi
+
+
 CMD="${bismark} ${nuc_ref} -1 ${R1} -2 ${R2} --multicore ${cpus} ${SUBSET_READS} \
     --un --ambiguous -o ${output_dir} --prefix b38_noMT --path_to_bowtie ${bowtie_path} \
     --samtools_path ${samtools_path} --temp_dir ${tmp_dir} > \