Commit dddf7af5 authored by TomKellyGenetics's avatar TomKellyGenetics
Browse files

update white lists for SCI-Seq (3-levels) and local test job

parent ead79e31

launch_universc.sh

+25 −8
Original line number Diff line number Diff line
@@ -1603,18 +1603,24 @@ else 
        elif [[ "$technology" == "sciseq2" ]]; then

             #generates all combinations of I1-I2-R1 barcodes

             if [[ ! -f ${whitelistdir}/sciseq2_barcode.txt ]]; then

                 join -j 9999 ${whitelistdir}/sci-seq3_i7_barcodes.txt ${whitelistdir}/sci-seq3_i5_barcodes.txt | sed "s/ //g" \

                 | join -j 9999 - ${whitelistdir}/sci-seq3_rt_barcodes.txt | sed "s/ //g" | awk '!a[$0]++'  > ${whitelistdir}/sciseq2_barcode.txt

                 join -j 9999 ${whitelistdir}/sci-seq3_i7_barcodes.txt ${whitelistdir}/sci-seq3_i5_barcodes.txt | sed "s/ //g" | \

                 join -j 9999 - ${whitelistdir}/sci-seq3_rt_barcodes.txt | sed "s/ //g" | awk '!a[$0]++'  > ${whitelistdir}/sciseq2_barcode.txt

             fi

        elif [[ "$technology" == "sciseq3" ]]; then

             if [[ ! -f ${whitelistdir}/sciseq3_barcode.txt ]]; then

                 #generates all combinations of I1-I2-R1 barcodes

                 join -j 9999 ${whitelistdir}/sci-seq3_i7_barcodes.txt ${whitelistdir}/sci-seq3_i5_barcodes.txt | sed "s/ //g" \

                 | join -j 9999 - ${whitelistdir}/sci-seq3_hp_barcodes.txt | sed "s/ //g" | join -j 9999 - ${whitelistdir}/sci-seq3_rt_barcodes.txt | sed "s/ //g" \

                 | awk '!a[$0]++'  > ${whitelistdir}/sciseq3_barcode.txt

                 join -j 9999 ${whitelistdir}/sci-seq3_i7_barcodes.txt ${whitelistdir}/sci-seq3_i5_barcodes.txt | sed "s/ //g" | \

                 join -j 9999 - ${whitelistdir}/sci-seq3_hp_barcodes.txt | sed "s/ //g" | join -j 9999 - ${whitelistdir}/sci-seq3_rt_barcodes.txt | sed "s/ //g" \

                 > ${whitelistdir}/sciseq3_barcode.txt

                 ## to filter unique lines: awk '!a[$0]++'  > ${whitelistdir}/sciseq3_barcode.txt

             fi

        else

            #generating permutations of ATCG of barcode length (non-standard evaluation required to run in script)

            if [[ ${barcodelength} -ge 12 ]]; then

                echo "  ... generating all permutations of A,T,C,G of length ${barcodelength}"

                echo "  WARNING: for large barcodes this could take a lot of time and memory"

                echo "  Please use a known barcode whitelist if possible"

            fi

            echo $(eval echo $(for ii in $(eval echo {1..${barcodelength}}); do echo "{A,T,C,G}"; done | tr "\n" " " | sed "s/ //g" | xargs -I {} echo {})) | sed 's/ /\n/g' | sort | uniq > ${barcodefile}

        fi

    fi
@@ -2405,6 +2411,7 @@ else 
        echo "  ...remove adapter for ${technology}"

        for convFile in "${convFiles[@]}"; do

            #remove adapter if detected (two-level indexing)

             ## TruSeq adapter: ACGACGCTCTTCCGATCT

            sed -E '

                /^ACGACGCTCTTCCGATCT/ {

                s/^ACGACGCTCTTCCGATCT(.{18})/\1/g
@@ -2439,9 +2446,19 @@ else 
        echo "  ...remove adapter for ${technology}"

        for convFile in "${convFiles[@]}"; do

            #remove adapter if detected (and keep hairpin/tn5 barcode)

            ## TruSeq adapter: ACGACGCTCTTCCGATCT

            sed -E '

                /^ACGACGCTCTTCCGATCT/ {

                s/^ACGACGCTCTTCCGATCT//g

                n

                n

                s/^.{18}//g

                }'  $convFile > ${crIN}/.temp

            mv ${crIN}/.temp $convFile

            #remove linker

            sed -E '

                /^ACGACGCTCTTCCGATCT(.{10})CAGAGC/ {

                s/^ACGACGCTCTTCCGATCT(.{10})CAGAGC(.{18})/\1\2/g

                /^(.{10})CAGAGC/ {

                s/^(.{10})CAGAGC(.{18})/\1\2/g

                n

                n

                s/^(.{26})(.{10})(.{6})(.{18})/\2\4/g

test/run_tests_sciseq_local.sh

0 → 100755
+62 −0
Original line number Diff line number Diff line 
#!/bin/bash



# run tests in universc directory (parent of test directory)

cd $(dirname ${BASH_SOURCE[0]})/..

pwd

##git pull --ff-only origin $(git branch --show-current) 



# used to export to PATH for testing on SGE server

export PATH=${HOME}/local/bin/cellranger-3.0.2:$PATH



cellrangerversion=`cellranger count --version | head -n 2 | tail -n 1 | cut -f2 -d'(' | cut -f1 -d')'`

cellrangerpath=`which cellranger`



# set up cellranger reference

if [[ ! -f test/cellranger_reference/cellranger-tiny-ref/3.0.0/star/SA ]] && [[ -f $(dirname $cellrangerpath)/cellranger-tiny-ref/3.0.0/star/SA ]]; then

    rsync $(dirname $cellrangerpath)/cellranger-tiny-ref/3.0.0/star/SA test/cellranger_reference/cellranger-tiny-ref/3.0.0/star/SA

fi

if [[ ! -f test/cellranger_reference/cellranger-tiny-ref/1.2.0/star/SA ]] && [[ -f $(dirname $cellrangerpath)/cellranger-tiny-ref/1.2.0/star/SA ]]; then

    rsync $(dirname $cellrangerpath)/cellranger-tiny-ref/1.2.0/star/SA test/cellranger_reference/cellranger-tiny-ref/1.2.0/star/SA

fi



##unpigz -k test/shared/sciseq-v3-test/SRR7827205*fastq.gz



if [ -f test/shared/sciseq-v3-test/SRR7827205_S1_L001_R1_001.fastq ]; then

    rename "s/_S1_L001/_L001/" test/shared/sciseq-v3-test/SRR7827205_S1_L001_R1_001.fastq*

fi

if [ -f test/shared/sciseq-v3-test/SRR7827205_S1_L001_R1_001.fastq ]; then

    rename "s/_S1_L001/_L001/" test/shared/sciseq-v3-test/SRR7827205_S1_L001_R1_001.fastq*

fi

if [ -f test/shared/sciseq-v3-test/SRR7827205_S2_L002_R1_001.fastq ]; then

    rename -n "s/_S2_L002/_L002/" test/shared/sciseq-v3-test/SRR7827205_S2_L002_R1_001.fastq*

fi

if [ -f test/shared/sciseq-v3-test/SRR7827205_S2_L002_R2_001.fastq ]; then

    rename -n "s/_S2_L002/_L002/" test/shared/sciseq-v3-test/SRR7827205_S2_L002_R2_001.fastq*

fi

if [ -d test-icell8-72618-KU812-2-lanes ];then

    rm -rf test-icell8-72618_KU812-2-lanes

fi



if [[ ! -f whitelists/sciseq3_barcode_test.txt ]]; then

    bash launch_universc.sh --id "test-sciseq" --technology "sciseq3" --setup --verbose

    grep "CCGAATCCGACTCCATCGA" whitelists/sciseq3_barcode.txt > whitelists/sciseq3_barcode_test.txt

fi



bash launch_universc.sh --id "test-sciseq" --technology "sciseq" \

 --reference "test/cellranger_reference/cellranger-tiny-ref/3.0.0" \

 --read1 "test/shared/sciseq-v3-test/SRR7827205_S1_R1.fastq.gz" \

 --read2 "test/shared/sciseq-v3-test/SRR7827205_S1_R2.fastq.gz" \

 --barcodefile whitelists/sciseq3_barcode_test.txt \

 --jobmode "local" --localcores 1 --verbose



bash launch_universc.sh --id "test-sciseq" --technology "sciseq" \

 --reference "test/cellranger_reference/cellranger-tiny-ref/3.0.0" \

 --read1 "test/shared/sciseq-v3-test/SRR7827205_S1_R1.fastq.gz" \

 --read2 "test/shared/sciseq-v3-test/SRR7827205_S1_R2.fastq.gz" \

 --jobmode "local" --localcores 1 



if [ -f test/shared/sciseq-v3-test/SRR7827205_S1_L001_R1_001.fastq.gz ]; then

    rename "s/_S1_L001/_S1/" test/shared/sciseq-v3-test/SRR7827 205_S1_L001_[IR][12]_001.fastq*

    rename "s/_001//" test/shared/sciseq-v3-test/SRR7827205_S1_[IR][12]_001.fastq*

fi

gzip test/shared/sciseq-v3-test/SRR7827205_S1_[IR][12].fastq

CRA Git | Maintained and supported by SUSTech CRA and CCSE