adding whippet support

core-w/2020-10-18_makeIndex.sh

@@ -57,7 +57,7 @@ do
   unset OPTIND
   unset OPTARG
 
-  while getopts a:b:o:t:v:m: options
+  while getopts a:n:c:o:t:v:m:w: options
   do
 
     case $options in
@@ -71,12 +71,17 @@ do
       export myAnnotation="${input}/$(ls ${OPTARG} | grep -e ".gtf")"
       ;;
 
-      b) # binary alignments
-      [ -d ${OPTARG} ] && export bamDir="${OPTARG}" && bamList=`ls -d ${OPTARG}/*.*` || die "cannot fetch input files" # get a list of files in the primary input
-      bamCount=$(echo $bamList | grep -c -e ".bam")
-      [ ${bamCount} -gt "0" ] || die "no binary alignments found in ${bamDir}"
+      n) # nacent RNA alignments
+      [ -d ${OPTARG} ] && export nbamDir="${OPTARG}" && nbamList=`ls -d ${OPTARG}/*.*` || die "cannot fetch input files" # get a list of files in the primary input
+      nbamCount=$(echo $nbamList | grep -c -e ".bam")
+      [ ${nbamCount} -gt "0" ] || die "no binary alignments found in ${nbamDir}"
       ;;
 
+      c) # cytoplasmic RNA alignments
+      [ -d ${OPTARG} ] && export cbamDir="${OPTARG}" && cbamList=`ls -d ${OPTARG}/*.*` || die "cannot fetch input files" # get a list of files in the primary input
+      cbamCount=$(echo $cbamList | grep -c -e ".bam")
+      [ ${cbamCount} -gt "0" ] || die "no binary alignments found in ${cbamDir}"
+      ;;
       o) # output directory
       export outputDirectory="${OPTARG}"
       mkdir -p ${outputDirectory} 2>/dev/null
@@ -108,6 +113,8 @@ do
       else die "could not parse the user provided runmode, ${OPTARG}"; fi
       ;;
 
+      w) # path to whippet index
+
     esac;
   done
   shift $((OPTIND-1))
@@ -123,7 +130,6 @@ done
 vsec "your annotation is ${myAnnotation}"
 vsec "your reference sequence is ${myReferenceSequence}"
 
-
 # validate the binary alignments
 [ "${bamCount}" -gt "0" ] || die "did not detect any bam files in the input"
 ssec "detected ${bamCount} bam files"
@@ -131,6 +137,7 @@ ssec "detected ${bamCount} bam files"
 # validate output directory
 [ -d "${outputDirectory}" ] || die "cannot access your output directory"
 vsec "Your output directory is ${outputDirectory}"
+export od=${outputDirectory}
 
 # validate threadCount
 ssec "proceeding with ${threadCount} threads"
@@ -141,5 +148,16 @@ ssec "your runmode is ${MODE}"
 
 vsec "completed housekeeping; parsing GTF"
 
-####################################################################################################
-####################################################################################################
+####################################
+
+# generate on the fly whippet index using mature bams
+
+# first, merge the mature bams
+matureSamMerge="${od}/matureSamMerge"; mkdir -p ${matureSamMerge} 2>/dev/null
+cd ${cbamDir} && samtools merge -@ ${threadCount} ${matureSamMerge}/merged.bam `ls *.bam` || die "cannot merge the cytoplasmic RNA alignments"
+samtools sort -@ ${threadCount} ${matureSamMerge}/merged.bam > ${matureSamMerge}/sorted.merged.bam || die "cannot sort your bam"
+samtools index -@ ${threadCount} ${matureSamMerge}/sorted.merged.bam || die "cannot index your merged bam"
+
+# next, generate on-the-fly whippet index
+
+  julia bin/whippet-index.jl