migrate index, mock UMI, and leading UMI to perl (efa7f0ca) · Commits · github_fork / Universc

launch_universc.sh

+9 −13

Original line number	Diff line number	Diff line
		@@ -1651,7 +1651,7 @@ if [[ "$technology" == "indrop-v3" ]] \|\| [[ "$technology" == "icell8-full-lengt
		if [[ $verbose ]]; then
		echo index of length $indexlength gives quality score $qualscores
		fi
		perl -pni -e "4~4s/^.*$/${qualscores}/g" $I1_file
		perl -pni -e "s/^.*$/${qualscores}/g if ($. % 4 == 0)" $I1_file
		#copies index 2 to next line (1st to 2nd) and deletes 3rd line
		cat $R1_file \| sed -E "s/ (.):(.):(.):(.)\+(.)$/ \1:\2:\3:\4+\5\n\5/g" \| perl -pn -e "3~5d" > $I2_file
		index2length=$(($(head $I2_file -n 2 \| tail -n 1 \| wc -c) - 1))
		@@ -1659,7 +1659,7 @@ if [[ "$technology" == "indrop-v3" ]] \|\| [[ "$technology" == "icell8-full-lengt
		if [[ $verbose ]]; then
		echo index2 of length $index2length gives quality score $qualscores2
		fi
		perl -pni -e "4~4s/^.*$/${qualscores2}/g" $I2_file
		perl -pni -e "s/^.*$/${qualscores2}/g if ($. % 4 == 0)" $I2_file
		index1+=("$I1_file")
		index2+=("$I2_file")
		done
		@@ -1701,7 +1701,7 @@ if [[ "$technology" == "quartz-seq" ]] \|\| [[ "$technology" == "ramda-seq" ]] \|\|
		if [[ $verbose ]]; then
		echo index of length $indexlength gives quality score $qualscores
		fi
		perl -pni -e "4~4s/^.*$/${qualscores}/g" $I1_file
		perl -pni -e "s/^.*$/${qualscores}/g if ($. % 4 == 0)" $I1_file
		index1+=("$I1_file")
		done
		if [[ $verbose ]]; then
		@@ -2946,7 +2946,7 @@ else
		echo " ... barcode and UMI swapped for ${technology}"
		for convFile in "${convFiles[@]}"; do
		#swap UMI and barcode
		sed -E '2~2s/(.{6})(.{6})/\2\1/' $convFile > ${crIN}/.temp
		perl -pni -e "s/(.{6})(.{6})/\1\2/' if ($. % 2 == 0)" $convFile
		mv ${crIN}/.temp $convFile
		done
		fi
		@@ -3257,8 +3257,7 @@ else
		mv ${crIN}/.temp $convFile
		#swap barcode and UMI
		echo " ... barcode and UMI swapped for ${technology}"
		sed -E '2~2s/(.{8})(.{10})/\2\1/' $convFile > ${crIN}/.temp
		mv ${crIN}/.temp $convFile
		perl -pni -e "s/(.{8})(.{10})/\1\2/' if ($. % 2 == 0)" $convFile

		read=$convFile
		convR1=$read
		@@ -3309,8 +3308,7 @@ else
		mv ${crIN}/.temp $convFile
		#swap barcode and UMI
		echo " ... barcode and UMI swapped for ${technology}"
		sed -E '2~2s/(.{10})(.{8})(.{10})/\1\3\2/' $convFile > ${crIN}/.temp
		mv ${crIN}/.temp $convFile
		perl -pni -e "s/(.{10})(.{8})(.{10})/\1\3\2/' if ($. % 2 == 0)" $convFile

		read=$convFile
		convR1=$read
		@@ -3716,11 +3714,9 @@ else
		#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 $convFile
		perl -pni -e "s/(.{$keeplength})(.*)/\1$toS\2/ if ($. % 4 == 2)" $convFile
		#Add n characters to the end of the quality
		sed -E "4~4s/(.{$keeplength})(.*)/\1$toQ\2/" $convFile > ${crIN}/.temp
		mv ${crIN}/.temp $convFile
		perl -pni -e "s/(.{$keeplength})(.*)/\1$toQ\2/ if ($. % 4 == 0)" $convFile
		echo " ${convFile} adjusted"
		done
		fi
		@@ -3732,7 +3728,7 @@ else
		#compute length of adjusted barcode + original UMI
		targetlength=`echo $((${barcode_default} + ${umi_default}))`
		#Remove n characters to the end of the sequence and quality score
		sed -E "2~2s/(.{$targetlength})(.{$umiadjust})(.*)/\1\3/" $convFile > ${crIN}/.temp
		perl -pni -e "s/(.{$targetlength})(.{$umiadjust})(.*)/\1\3/ if ($. % 2 == 0)" $convFile
		mv ${crIN}/.temp $convFile
		echo " ${convFile} adjusted"
		done

test/run_tests_icell8_local.sh

+6 −6

Original line number	Diff line number	Diff line
		@@ -38,12 +38,12 @@ fi
		if [ -d test-icell8-72618-KU812-2-lanes-non-umi ];then
		rm -rf test-icell8-72618_KU812-2-lanes-non-umi
		fi
		bash launch_universc.sh --id "test-icell8-72618_KU812-2-lanes-non-umi" --technology "iCell8" \
		--reference "test/cellranger_reference/cellranger-tiny-ref/3.0.0" \
		--read1 "test/shared/icell8-test/72618_KU812_L001_R1_001.fastq" "test/shared/icell8-test/72618_KU812_L002_R1_001.fastq" \
		--read2 "test/shared/icell8-test/72618_KU812_L001_R2_001.fastq" "test/shared/icell8-test/72618_KU812_L002_R2_001.fastq" \
		--barcodefile "test/shared/icell8-test/WellList.txt" \
		--per-cell-data --jobmode "local" --localcores 2 --localmem 4 --non-umi
		#bash launch_universc.sh --id "test-icell8-72618_KU812-2-lanes-non-umi" --technology "iCell8" \
		# --reference "test/cellranger_reference/cellranger-tiny-ref/3.0.0" \
		# --read1 "test/shared/icell8-test/72618_KU812_L001_R1_001.fastq" "test/shared/icell8-test/72618_KU812_L002_R1_001.fastq" \
		# --read2 "test/shared/icell8-test/72618_KU812_L001_R2_001.fastq" "test/shared/icell8-test/72618_KU812_L002_R2_001.fastq" \
		# --barcodefile "test/shared/icell8-test/WellList.txt" \
		# --per-cell-data --jobmode "local" --localcores 2 --localmem 4 --non-umi

		if [ -f test/shared/icell8-test/72618_KU812_S1_L001_R1_001.fastq ]; then
		rename "s/_S1_L001/_L001/" test/shared/icell8-test/72618_KU812_S1_L001_R1_001.fastq*

Admin message