Demux script for C4W17ACXX

C4W17ACXX_demux.sh

#!/bin/bash

NAME="C4W17ACXX"
EMAIL='lg1883@nyu.edu'

FLOWCELL_PATH="/scratch/gencore/hiseq-700911/140710_SN911_0221_AC4W17ACXX"
OUTPUT_PATH="/scratch/gencore/manual/140710_SN911_0221_AC4W17ACXX"

CONFIGURE_BCL2FASTQ="\
--fastq-cluster-count 0 \
--use-bases-mask \"Y*,I6n\" \
--no-eamss \
--mismatches 1"

SAMPLESHEET_CONTENT="\
C4W17ACXX,2,Undetermined,,Undetermined,,N,,lhw235@nyu.edu,Undetermined
C4W17ACXX,2,LW175,,GTCCGC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW176,,GTGAAA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW177,,GTGGCC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW178,,GTTTCG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW179,,CGTACG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW180,,GAGTGG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW181,,GGTAGC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW182,,ATCACG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW183,,CGATGT,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW184,,TTAGGC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW185,,TGACCA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW186,,ACAGTG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW187,,GCCAAT,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW188,,CAGATC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW189,,ACTTGA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW190,,GATCAG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW191,,TAGCTT,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW192,,GGCTAC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW193,,CTTGTA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,2,LW194,,AGTCAA,,N,,lhw235@nyu.edu,Condensins
C4W17ACXX,2,LW195,,AGTTCC,,N,,lhw235@nyu.edu,Condensins
C4W17ACXX,2,LW196,,ATGTCA,,N,,lhw235@nyu.edu,Seaurchin
C4W17ACXX,3,Undetermined,,Undetermined,,N,,lhw235@nyu.edu,Undetermined
C4W17ACXX,3,LW197,,CCGTCC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW198,,GTAGAG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW199,,GTCCGC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW200,,GTGAAA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW201,,GTGGCC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW202,,GTTTCG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW203,,CGTACG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW204,,GAGTGG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW205,,GGTAGC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW206,,ATCACG,,N,,lhw235@nyu.edu,Seaurchin
C4W17ACXX,3,LW207,,CGATGT,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW208,,TTAGGC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW209,,TGACCA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW210,,ACAGTG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW211,,GCCAAT,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW212,,CAGATC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW213,,ACTTGA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW214,,GATCAG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW215,,TAGCTT,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW216,,GGCTAC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW217,,CTTGTA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW218,,AGTCAA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW219,,AGTTCC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,3,LW220,,ATGTCA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,Undetermined,,Undetermined,,N,,lhw235@nyu.edu,Undetermined
C4W17ACXX,4,LW221,,CCGTCC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW222,,GTAGAG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW223,,GTCCGC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW224,,GTGAAA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW225,,GTGGCC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW226,,GTTTCG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW227,,CGTACG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW228,,GAGTGG,,N,,lhw235@nyu.edu,Seaurchin
C4W17ACXX,4,LW229,,GGTAGC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW230,,ATCACG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW231,,CGATGT,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW232,,TTAGGC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW233,,TGACCA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW234,,ACAGTG,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW235,,GCCAAT,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW236,,CAGATC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW237,,ACTTGA,,N,,lhw235@nyu.edu,Condensins
C4W17ACXX,4,LW238,,GATCAG,,N,,lhw235@nyu.edu,Condensins
C4W17ACXX,4,LW239,,TAGCTT,,N,,lhw235@nyu.edu,Condensins
C4W17ACXX,4,LW240,,GGCTAC,,N,,lhw235@nyu.edu,Condensins
C4W17ACXX,4,LW241,,CTTGTA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW242,,AGTCAA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW243,,AGTTCC,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,4,LW244,,ATGTCA,,N,,lhw235@nyu.edu,DosageComp
C4W17ACXX,5,Undetermined,,Undetermined,,N,,nz375@nyu.edu,Undetermined
C4W17ACXX,5,SoilN_1,,CGATGT,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,5,SoilF_1,,TGACCA,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,5,DGY952_1,,ACAGTG,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,5,DGY955_1,,GCCAAT,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,5,SoilF2Pool_1,,CAGATC,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,6,Undetermined,,Undetermined,,N,,nz375@nyu.edu,Undetermined
C4W17ACXX,6,DGY1162_1,,CTTGTA,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,6,DGY1163_1,,ATCACG,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,6,DGY1166_1,,TTAGGC,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,6,DGY1167_1,,ACTTGA,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,6,DGY1179_1,,GATCAG,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,6,DGY1180_1,,TAGCTT,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,6,DGY1181_1,,GGCTAC,,N,,nz375@nyu.edu,SoilStrains
C4W17ACXX,7,Undetermined,,Undetermined,,N,,sv49@nyu.edu,Undetermined
C4W17ACXX,7,SV16,,ACAGTG,,N,,sv49@nyu.edu,NILProgramming
C4W17ACXX,7,SV17,,CAGATC,,N,,sv49@nyu.edu,NILProgramming
C4W17ACXX,7,SV18,,GCCAAT,,N,,sv49@nyu.edu,NILProgramming
C4W17ACXX,7,SV19,,TGACCA,,N,,sv49@nyu.edu,NILProgramming
C4W17ACXX,7,SV20,,CGATGT,,N,,sv49@nyu.edu,NILProgramming
C4W17ACXX,8,Undetermined,,Undetermined,,N,,sv49@nyu.edu,Undetermined
C4W17ACXX,8,SV21,,TGACCA,,N,,sv49@nyu.edu,NILProgramming
C4W17ACXX,8,SV22,,GCCAAT,,N,,sv49@nyu.edu,NILProgramming
C4W17ACXX,8,SV23,,ACAGTG,,N,,sv49@nyu.edu,NILProgramming
C4W17ACXX,8,SV24,,CGATGT,,N,,sv49@nyu.edu,NILProgramming
C4W17ACXX,8,SV25,,CAGATC,,N,,sv49@nyu.edu,NILProgramming
"

CORES=12
BCL_TO_FASTQ_CMD="/share/apps/bcl2fastq/1.8.4/gnu/bin/configureBclToFastq.pl"
SAMPLE_SHEET_HEAD="FCID,Lane,SampleID,SampleRef,Index,Description,Control,Recipe,Operator,SampleProject"

BCL2FASTQ_PATH="$OUTPUT_PATH/bcl2fastq"
FASTQC_PATH="$OUTPUT_PATH/fastqc"
PBS_PATH="$OUTPUT_PATH/bcl2fastq.pbs"
SAMPLESHEET_PATH="$OUTPUT_PATH/SampleSheet.csv"

PBS_HEAD="#!/bin/sh
#PBS -V
#PBS -l nodes=1:ppn=$CORES,walltime=48:00:00
#PBS -M $EMAIL
#PBS -m abe
#PBS -q s48
#PBS -l mem=16GB"

make_bcl2fastq_pbs() {
echo -e "\
$PBS_HEAD

$BCL_TO_FASTQ_CMD \
--input-dir $FLOWCELL_PATH/Data/Intensities/BaseCalls  \
--output-dir $BCL2FASTQ_PATH/Unaligned \
--sample-sheet $SAMPLESHEET_PATH \
$CONFIGURE_BCL2FASTQ;

cd $BCL2FASTQ_PATH/Unaligned;

make -j$CORES;
result=\$?;

# load the fastqc module
module load fastqc

# run fastqc on each of the fastq files
for fastq in $BCL2FASTQ_PATH/Unaligned/Project_*/*/*.fastq.gz
do
    fastqc --noextract --threads $CORES --outdir $FASTQC_PATH \$fastq;
done

# make a tarball out of all fastqc reports
cd $OUTPUT_PATH && tar -cjf $NAME.tar.bz2 fastqc

exit \$result;" > $PBS_PATH
}

make_sample_sheet(){
    echo -e "$SAMPLE_SHEET_HEAD\n$SAMPLESHEET_CONTENT" > $SAMPLESHEET_PATH
}

make_output_directories(){
    rm -rf "$OUTPUT_PATH"
    mkdir -p "$OUTPUT_PATH"
    mkdir -p "$BCL2FASTQ_PATH"
    mkdir -p "$FASTQC_PATH"
}

make_output_directories;
make_sample_sheet;
make_bcl2fastq_pbs;

qsub -N "B2F_$NAME" $PBS_PATH