Changeset 5836

Timestamp:

Feb 24, 2020, 1:38:11 PM (3 years ago)

Author:

Nicklas Nordborg

Message:

References #1218: Implement MIPs alignment

Added functions for converting FASTQ to BAM and annotating with UMI information.

File:

• other/pipeline/trunk/mips_functions.sh (modified) (4 diffs)

other/pipeline/trunk/mips_functions.sh

@@ -61 +61 @@
 # Run two Trimmomatic steps. First trim adapters and then quality
 # Parameters:
-#   $1: Prefix to FASTQ files "_R?.fastq.gz" is added automatilly
-#   $2: Prefix for generated files
+#   $1: Prefix to FASTQ files "_R?.fastq.gz" is added automatically
+#   $2: Prefix for generated files (counter)
 # Global parameters:
 #   $JAVA: Path to Java exectable
@@ -81 +81 @@
     fastq/${prefix}_R1.fastq.gz fastq/${prefix}_R2.fastq.gz \
     tmp/${n}.adaptrim.1.fastq tmp/${n}.adaptrim.u.1.fastq tmp/${n}.adaptrim.2.fastq tmp/${n}.adaptrim.u.2.fastq \
-    ${TrimOptionsAdapter}
+    ${TrimOptionsAdapter} \
     >> trimmomatic.out
     
@@ -97 +97 @@
     tmp/${n}.adaptrim.1.fastq tmp/${n}.adaptrim.2.fastq \
     tmp/${n}.qualtrim.1.fastq tmp/${n}.qualtrim.u.1.fastq tmp/${n}.qualtrim.2.fastq tmp/${n}.qualtrim.u.2.fastq \
-    ${TrimOptionsQual}
+    ${TrimOptionsQual} \
     >> trimmomatic.out
     
@@ -123 +123 @@
 }
 
+# Read RG file and convert to Picard2 syntax.
+# Parameters:
+#   $1: Path to RG file
+# Returns:
+#   Converted RG value
+function read_and_fix_rg {
+  local rg=$1
+  
+  read -r DATA < ${rg}
+  
+  DATA=${DATA/RG=/-READ_GROUP_NAME }
+  DATA=${DATA/SM=/-SAMPLE_NAME }
+  DATA=${DATA/LB=/-LIBRARY_NAME }
+  DATA=${DATA/PU=/-PLATFORM_UNIT }
+  DATA=${DATA/DT=/-RUN_DATE }
+  DATA=${DATA/PL=/-PLATFORM  }
+  DATA=${DATA/CN=/-SEQUENCING_CENTER }
+  
+  echo ${DATA}
+}
 
+# Run Picard to convert FASTQ files to BAM file
+# Parameters:
+#   $1: Prefix to RG file ".rg" is added automatically
+#   $2: Prefix for generated files (counter)
+function fastq_to_bam {
+  local prefix=$1
+  local n=$2
+  
+  if [ ! -f tmp/${n}.qualtrim.1.fastq ]; then
+    return
+  fi
+  
+  local rg=$(read_and_fix_rg "fastq/${prefix}.rg")
+  
+  ./stdwrap.sh ./picard2 FastqToSam \
+    -SORT_ORDER queryname \
+    -QUALITY_FORMAT Standard \
+    ${rg} \
+    -FASTQ tmp/${n}.qualtrim.1.fastq \
+    -FASTQ2 tmp/${n}.qualtrim.2.fastq \
+    -OUTPUT tmp/${n}.u.bam \
+    >> fastqtosam.out
+}
+
+# Run FgBio to annotate the BAM file with UMI information
+# Parameters:
+#   $1: Prefix to UMI file "_UMI.fastq.gz" is added automatically
+#   $2: Prefix for generated files (counter)
+# Global parameters:
+#   $JAVA: Path to Java exectable
+#   $FgBioJAR: Path to FgBio JAR file
+function annotate_umi {
+  local prefix=$1
+  local n=$2
+  
+  if [ ! -f tmp/${n}.u.bam ]; then
+    return
+  fi
+  
+  ./stdwrap.sh ${JAVA} -jar ${FgBioJAR} AnnotateBamWithUmis \
+    --fail-fast true -t RX \
+    -i tmp/${n}.u.bam \
+    -f fastq/${prefix}_UMI.fastq.gz \
+    -o tmp/${n}.umi.bam \
+    >> annotatebam.out
+
+}
+
+