added rm encode tmp

utility/encode_step2_rt_rl_correction_and_fq_transfering.py

@@ -63,7 +63,7 @@ def main():
                                 subprocess.call(cmd_wget,shell=True)
                                 origin = os.path.join(encodetmdir,fq_url.rsplit('/',1)[1])                           
                                 newname = os.path.join(fqdir,fields[0]+'.fastq.gz')
-                                cmd_ln = 'ln -s '+ origin +' ' + newname
+                                cmd_ln = 'ln -fs '+ origin +' ' + newname
                                 print(cmd_ln)
                                 subprocess.call(cmd_ln,shell=True)
 
@@ -77,7 +77,7 @@ def main():
                                 subprocess.call(cmd_wget,shell=True)
                                 origin = os.path.join(encodetmdir,fq_url.rsplit('/',1)[1])                           
                                 newname = os.path.join(fqdir,fields[0]+'_R'+str(r1_or_r2)+'.fastq.gz')
-                                cmd_ln = 'ln -s '+ origin +' ' + newname
+                                cmd_ln = 'ln -sf '+ origin +' ' + newname
                                 print(cmd_ln)
                                 subprocess.call(cmd_ln,shell=True)
                 seq_info.read_length = str(read_length)

utility/runSetQC.sh

@@ -70,3 +70,21 @@ ssh [email protected] $cmd2
 #ver=$(ssh [email protected] $cmd)
 #url="http://epigenomics.sdsc.edu:8088/${SET_ID}/$(cat ${SETQC_FILE/.txt/.rstr.txt})/setQC_report.html" 
 #python updateLibrariesSetQC.py -s '3' -url $url -v $ver -id $SET_ID
+
+
+##################################################
+##  Step 4. (optional) delete ENCODE raw fastq files
+##################################################
+SEQ_DIR='/projects/ps-epigen/seqdata/'
+libs=($(awk -v FS='\t' '(NR>1){print $1}' $STATUS_FILE))
+for lib in ${libs[@]}
+do
+    if [ $lib = "ENCODE_"* ]
+    then
+        echo $lib
+        for lib_link in ${SEQ_DIR}/${lib}.fastq.gz ${SEQ_DIR}/${lib}_R1.fastq.gz ${SEQ_DIR}/${lib}_R2.fastq.gz
+        do
+            [[ ! -z ${lib_link} ]] && rm "$(readlink -f $lib_link)"
+        done
+    fi
+done

utility/runSetQC_chipseq.sh

@@ -23,7 +23,6 @@ groups=($(awk '(NR>1){print $2}' $STATUS_FILE|uniq))
 n_groups=${#groups[@]}
 n_libs=$(awk -v FS='\t' 'BEGIN{n=0};{if(NR>1&&$6=="No") n=n+1}END{print n}' $STATUS_FILE)
 
-
 ## determine which pipeline to run
 if [ $(grep -c True $STATUS_FILE) -eq 0 ] 
 then
@@ -86,3 +85,20 @@ ssh [email protected] $cmd2
 
 
 #python updateLibrariesSetQC.py -s '3' -url $url -v $ver -id $SET_ID
+
+##################################################
+##  Step 4. (optional) delete ENCODE raw fastq files
+##################################################
+SEQ_DIR='/projects/ps-epigen/seqdata/'
+libs=($(awk -v FS='\t' '(NR>1){print $1}' $STATUS_FILE))
+for lib in ${libs[@]}
+do
+    if [ $lib = "ENCODE_"* ]
+    then
+        echo $lib
+        for lib_link in ${SEQ_DIR}/${lib}.fastq.gz ${SEQ_DIR}/${lib}_R1.fastq.gz ${SEQ_DIR}/${lib}_R2.fastq.gz
+        do
+            [[ ! -z ${lib_link} ]] && rm "$(readlink -f $lib_link)"
+        done
+    fi
+done