From aae31bf93f5829b7e8eb496037ff8dd441a486f7 Mon Sep 17 00:00:00 2001
From: Smith Nicholas <smith@in.tum.de>
Date: Fri, 26 Aug 2022 09:08:52 +0200
Subject: [PATCH] check for empty Platform (PL) in bam RG

---
 .../rvc-pipeline/GATK_BASH/changeHeader.sh    | 45 +++++++++++--------
 1 file changed, 27 insertions(+), 18 deletions(-)

diff --git a/drop/modules/rvc-pipeline/GATK_BASH/changeHeader.sh b/drop/modules/rvc-pipeline/GATK_BASH/changeHeader.sh
index 75e746a9..3e97d2db 100755
--- a/drop/modules/rvc-pipeline/GATK_BASH/changeHeader.sh
+++ b/drop/modules/rvc-pipeline/GATK_BASH/changeHeader.sh
@@ -18,41 +18,50 @@ output_bai=$6
 output_newHeader=$7
 
 
+SM_internalHeader=""
+PL_internalHeader=""
 samtools view -H $input_bam |grep "^@RG" |grep SM |head -1|
 while read header ; do
     for i in $header; do
         if [[ $i == "SM:"* ]]; then
-            internalHeader=${i:3}
-            break
-        else
-            internalHeader=""
+            SM_internalHeader=${i:3}
+        fi
+        if [[ $i == "PL:"* ]]; then
+            PL_internalHeader=${i:3}
         fi
     done
 
-    if [[ $internalHeader == $sample ]]; then
-        echo "Internal Header $internalHeader matches $sample" |tee $log
-        echo "Internal Header is designated: $internalHeader"  |tee -a $log
+    if [[ $SM_internalHeader == $sample && $PL_internalHeader != "" ]]; then
+        echo "Internal Header $SM_internalHeader matches $sample" |tee $log
+        echo "Internal Header is designated: $SM_internalHeader"  |tee -a $log
         echo "SampleID is $sample" |tee -a $log
         ln -f  $input_bam $output_bam
         ln -f  $input_bai $output_bai
         echo "Done Linking files"
 		samtools view -H $input_bam > $output_newHeader
     else
-        echo "WARNING"
-        echo "Internal Header is designated: $internalHeader" |tee $log
-        echo "SampleID is $sample"  |tee -a $log
-        echo "Forcing $internalHeader to match $sample" |tee -a $log
-
-        samtools view -H $input_bam > $output_newHeader
-        echo $output_newHeader
+        if [[ $SM_internalHeader != $sample ]]; then
+            echo "WARNING"
+            echo "Internal Header is designated: $SM_internalHeader" |tee $log
+            echo "SampleID is $sample"  |tee -a $log
+            echo "Forcing $SM_internalHeader to match $sample" |tee -a $log
 
-        # sed using regEx in place substitiute 'SM:' followed by any thing that isn't tab or newLine. and then replace it with the sampleID and the delimiter (tab or newLine) that matched in the 1st expression.
-        sed -E -i "s/(SM:[^\t|\n]*)(\t|\n*)/SM:${sample}\2/" ${output_newHeader}
+            # sed using regEx in place substitiute 'SM:' followed by any thing that isn't tab or newLine. and then replace it with the sampleID and the delimiter (tab or newLine) that matched in the 1st expression.
+            samtools view -H $input_bam > $output_newHeader
+            sed -E -i "s/(SM:[^\t|\n]*)(\t|\n*)/SM:${sample}\2/" ${output_newHeader}
+        fi
+        if [[ $PL_internalHeader == "" ]]; then
+            echo "WARNING"
+            echo "Internal PL Header is not designated: $PL_internalHeader" |tee $log
+            echo "PL cannot be empty. Using a Dummy: dummyPL "  |tee -a $log
+            echo "Forcing $PL_internalHeader to dummyPL" |tee -a $log
 
+            # sed using regEx in place substitiute '@RG' with @RG\tPL:dummyPL
+            samtools view -H $input_bam > $output_newHeader
+            sed -E -i "s/^@RG/@RG\tPL:dummyPL/" ${output_newHeader}
+        fi
         samtools reheader $output_newHeader $input_bam > $output_bam
-
         samtools index -b $output_bam
-
     fi
     echo "new header can be found here:$output_newHeader" |tee -a $log
 done