replace UMI with mock (removes hardcoded barcode length) by non-standard eval (d0bb5644) · Commits · github_fork / Universc

launch_universc.sh

+29 −44

Original line number	Original line	Diff line number	Diff line
	@@ -3215,26 +3215,6 @@ else
	fi		fi
	fi		fi

	#convert UMI
	echo " adjusting UMIs of R1 files"
	# check if original UMI is shorter than default
	if [[ 0 -gt $umiadjust ]]; then
	for convFile in "${convFiles[@]}"; do
	echo " handling $convFile ..."
	toS=`printf '%0.sA' $(seq 1 $(($umiadjust * -1)))`
	toQ=`printf '%0.sI' $(seq 1 $(($umiadjust * -1)))`
	#compute length of adjusted barcode + original UMI
	keeplength=`echo $((${barcode_default}+${umi_default}-($umiadjust * -1)))`
	#Add n characters to the end of the sequence
	sed -E "2~4s/(.{$keeplength})(.*)/\1$toS\2/" $convFile > ${crIN}/.temp
	mv ${crIN}/.temp $convFile
	#Add n characters to the end of the quality
	sed -E "4~4s/(.{$keeplength})(.*)/\1$toQ\2/" $convFile > ${crIN}/.temp
	mv ${crIN}/.temp $convFile
	echo " ${convFile} adjusted"
	done
	fi

	#replace UMI with mock UMI to count reads (for technologies not already containing mock UMI)		#replace UMI with mock UMI to count reads (for technologies not already containing mock UMI)
	if [[ $technology != "icell8" ]] && [[ $technology != "ramda-seq" ]] && [[ $technology != "quartz-seq" ]] && [[ $technology != "smartseq" ]] && [[ $technology != "smartseq2" ]] && [[ $technology != "strt-seq" ]]; then		if [[ $technology != "icell8" ]] && [[ $technology != "ramda-seq" ]] && [[ $technology != "quartz-seq" ]] && [[ $technology != "smartseq" ]] && [[ $technology != "smartseq2" ]] && [[ $technology != "strt-seq" ]]; then
	if [[ $nonUMI ]]; then		if [[ $nonUMI ]]; then
	@@ -3244,35 +3224,20 @@ else

	for convFile in "${convFiles[@]}"; do		for convFile in "${convFiles[@]}"; do
	convR1=$convFile		convR1=$convFile
	#remove inflated umi (to replace with mock and count as reads)		#remove inflated umi (to replace with mock and count as reads) by non-standard evaluation to depend on variable umi-length
	sed -E "		cmd=$(echo 'sed -E "2~4s/\^(.{'$barcodelength'})(.{'${umilength}'})(.\*)$/\1\3/" '$convFile' > '${crIN}'/.temp')
	/^(.{16})(.{10})(.*)/ {
	s/^(.{16})(.{10})(.*)/\1\3/g
	n
	n
	s/^(.{16})(.{10})(.*)/\1\3/g
	}" $convFile > ${crIN}/.temp\n
	mv ${crIN}/.temp $convFile'
	if [[ $chemistry == "SC3Pv3" ]]; then
	chemistry="SC3Pv2"
	fi
	#cmd=$(echo 'sed -E "
	# /^(.{'$barcodelength'})(.{'${umilength}'})(.*)/ {
	# s/^(.{'$barcodelength'})(.{'${umilength}'})(.*)/\1\3/g
	# n
	# n
	# s/^(.{'$barcodelength'})(.{'${umilength}'})(.*)/\1\3/g
	# }" $convFile > ${crIN}/.temp\n
	# mv ${crIN}/.temp $convFile')
	if [[ $verbose ]]; then		if [[ $verbose ]]; then
	echo technology $technology		echo technology $technology
	echo barcode: $barcodelength		echo barcode: $barcodelength
	echo umi: $umilength		echo umi: $umilength
	# echo $cmd		echo $cmd
	fi		fi
	#eval $cmd		eval $cmd
			mv ${crIN}/.temp $convFile

	# add mock UMI (count reads instead of UMI) barcodelength=16, umi_default=10		if [[ $chemistry == "SC3Pv3" ]]; then
			chemistry="SC3Pv2"
			fi
	perl sub/AddMockUMI.pl --fastq=${convR1} --out_dir $crIN --head_length=$barcodelength --umi_length=$umi_default		perl sub/AddMockUMI.pl --fastq=${convR1} --out_dir $crIN --head_length=$barcodelength --umi_length=$umi_default
	umilength=$umi_default		umilength=$umi_default
	umiadjust=0		umiadjust=0
	@@ -3287,6 +3252,26 @@ else
	fi		fi
	fi		fi

			#convert UMI
			echo " adjusting UMIs of R1 files"
			# check if original UMI is shorter than default
			if [[ 0 -gt $umiadjust ]]; then
			for convFile in "${convFiles[@]}"; do
			echo " handling $convFile ..."
			toS=`printf '%0.sA' $(seq 1 $(($umiadjust * -1)))`
			toQ=`printf '%0.sI' $(seq 1 $(($umiadjust * -1)))`
			#compute length of adjusted barcode + original UMI
			keeplength=`echo $((${barcode_default}+${umi_default}-($umiadjust * -1)))`
			#Add n characters to the end of the sequence
			sed -E "2~4s/(.{$keeplength})(.*)/\1$toS\2/" $convFile > ${crIN}/.temp
			mv ${crIN}/.temp $convFile
			#Add n characters to the end of the quality
			sed -E "4~4s/(.{$keeplength})(.*)/\1$toQ\2/" $convFile > ${crIN}/.temp
			mv ${crIN}/.temp $convFile
			echo " ${convFile} adjusted"
			done
			fi

	# check if original UMI is longer than default		# check if original UMI is longer than default
	if [[ 0 -lt $umiadjust ]]; then		if [[ 0 -lt $umiadjust ]]; then
	for convFile in "${convFiles[@]}"; do		for convFile in "${convFiles[@]}"; do

Admin message