| 1 | #Set parameters and values |
|---|
| 2 | GSAPUBFTP_PATH="bundle/2.8/b37" |
|---|
| 3 | ROOT="/volume/" |
|---|
| 4 | RESDIR="$ROOT/resources/b37/" |
|---|
| 5 | TMPDIR="$ROOT/tmp" |
|---|
| 6 | BWAVERSION="0.7.12-goolf-1.7.20" |
|---|
| 7 | |
|---|
| 8 | |
|---|
| 9 | ###Create tmp directory and cd to it |
|---|
| 10 | mkdir -p $TMPDIR |
|---|
| 11 | cd $TMPDIR |
|---|
| 12 | |
|---|
| 13 | |
|---|
| 14 | ###wget resources |
|---|
| 15 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/human_g1k_v37.dict.gz |
|---|
| 16 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/human_g1k_v37.dict.gz.md5 |
|---|
| 17 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/human_g1k_v37.fasta.gz |
|---|
| 18 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/human_g1k_v37.fasta.gz.md5 |
|---|
| 19 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/human_g1k_v37.fasta.fai.gz |
|---|
| 20 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/human_g1k_v37.fasta.fai.gz.md5 |
|---|
| 21 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/Mills_and_1000G_gold_standard.indels.b37.vcf.gz |
|---|
| 22 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/Mills_and_1000G_gold_standard.indels.b37.vcf.gz.md5 |
|---|
| 23 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/Mills_and_1000G_gold_standard.indels.b37.vcf.idx.gz |
|---|
| 24 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/Mills_and_1000G_gold_standard.indels.b37.vcf.idx.gz.md5 |
|---|
| 25 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/dbsnp_138.b37.vcf.gz |
|---|
| 26 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/dbsnp_138.b37.vcf.gz.md5 |
|---|
| 27 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/dbsnp_138.b37.vcf.idx.gz |
|---|
| 28 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/dbsnp_138.b37.vcf.idx.gz.md5 |
|---|
| 29 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/dbsnp_138.b37.excluding_sites_after_129.vcf.gz |
|---|
| 30 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/dbsnp_138.b37.excluding_sites_after_129.vcf.gz.md5 |
|---|
| 31 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/dbsnp_138.b37.excluding_sites_after_129.vcf.idx.gz |
|---|
| 32 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/dbsnp_138.b37.excluding_sites_after_129.vcf.idx.gz.md5 |
|---|
| 33 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/1000G_phase1.indels.b37.vcf.gz |
|---|
| 34 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/1000G_phase1.indels.b37.vcf.gz.md5 |
|---|
| 35 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/1000G_phase1.indels.b37.vcf.idx.gz |
|---|
| 36 | wget ftp://gsapubftp-anonymous@ftp.broadinstitute.org//$GSAPUBFTP_PATH/1000G_phase1.indels.b37.vcf.idx.gz.md5 |
|---|
| 37 | wget https://molgenis26.target.rug.nl/downloads/intervals/testKit1_baits_b37_human_g1k_v37.bed |
|---|
| 38 | wget https://molgenis26.target.rug.nl/downloads/intervals/testKit1_baits_b37_human_g1k_v37.interval_list |
|---|
| 39 | wget https://molgenis26.target.rug.nl/downloads/intervals/testKit1_exons_b37_human_g1k_v37.bed |
|---|
| 40 | wget https://molgenis26.target.rug.nl/downloads/intervals/testKit1_exons_b37_human_g1k_v37.interval_list |
|---|
| 41 | wget https://molgenis26.target.rug.nl/downloads/intervals/1000G_phase1.indels_Mills_and_1000G_gold_standard.indels.b37.human_g1k_v37.chr1.intervals |
|---|
| 42 | |
|---|
| 43 | |
|---|
| 44 | ###Generate correct md5sum files |
|---|
| 45 | for i in $( ls *.md5 ); |
|---|
| 46 | do |
|---|
| 47 | echo "Generating correct md5sum file for $i"; |
|---|
| 48 | |
|---|
| 49 | SUM=`awk '{print $1}' $i`; |
|---|
| 50 | FILENAME=`cat $i | xargs -0 basename`; |
|---|
| 51 | |
|---|
| 52 | echo -e "$SUM $FILENAME" > $i.proper.md5; |
|---|
| 53 | |
|---|
| 54 | rm $i; |
|---|
| 55 | |
|---|
| 56 | done |
|---|
| 57 | |
|---|
| 58 | |
|---|
| 59 | ###Check md5sums generated files |
|---|
| 60 | for i in $( ls *.proper.md5 ); |
|---|
| 61 | do |
|---|
| 62 | echo "Checking md5sum for file $i"; |
|---|
| 63 | |
|---|
| 64 | md5sum -c $i; |
|---|
| 65 | |
|---|
| 66 | done |
|---|
| 67 | |
|---|
| 68 | |
|---|
| 69 | ###Unzip all *.gz files |
|---|
| 70 | for i in $( ls *.gz ); |
|---|
| 71 | do |
|---|
| 72 | echo "Gunzipping file $i"; |
|---|
| 73 | |
|---|
| 74 | gunzip $i; |
|---|
| 75 | |
|---|
| 76 | done |
|---|
| 77 | |
|---|
| 78 | |
|---|
| 79 | ###Index reference genome using BWA |
|---|
| 80 | module use /srv/molgenis/.local/easybuild/modules/all/ |
|---|
| 81 | module avail |
|---|
| 82 | module load BWA/$BWAVERSION |
|---|
| 83 | |
|---|
| 84 | bwa index -p human_g1k_v37.fasta -a bwtsw human_g1k_v37.fasta |
|---|
| 85 | |
|---|
| 86 | |
|---|
| 87 | ###Create folder structure for resources |
|---|
| 88 | mkdir -p $RESDIR/indices/ |
|---|
| 89 | mkdir -p $RESDIR/intervals/ |
|---|
| 90 | mkdir -p $RESDIR/snp/dbSNP/ |
|---|
| 91 | mkdir -p $RESDIR/snp/1000G/ |
|---|
| 92 | mkdir -p $RESDIR/sv/1000G |
|---|
| 93 | |
|---|
| 94 | |
|---|
| 95 | ###Move all resources to correct folder |
|---|
| 96 | mv human_g1k_v37.dict $RESDIR/indices/ |
|---|
| 97 | mv human_g1k_v37.fasta $RESDIR/indices/ |
|---|
| 98 | mv human_g1k_v37.fasta.amb $RESDIR/indices/ |
|---|
| 99 | mv human_g1k_v37.fasta.ann $RESDIR/indices/ |
|---|
| 100 | mv human_g1k_v37.fasta.bwt $RESDIR/indices/ |
|---|
| 101 | mv human_g1k_v37.fasta.pac $RESDIR/indices/ |
|---|
| 102 | mv human_g1k_v37.fasta.sa $RESDIR/indices/ |
|---|
| 103 | mv human_g1k_v37.fasta.fai $RESDIR/indices/ |
|---|
| 104 | mv Mills_and_1000G_gold_standard.indels.b37.vcf $RESDIR/sv/1000G |
|---|
| 105 | mv Mills_and_1000G_gold_standard.indels.b37.vcf.idx $RESDIR/sv/1000G |
|---|
| 106 | mv dbsnp_138.b37.vcf $RESDIR/snp/dbSNP/ |
|---|
| 107 | mv dbsnp_138.b37.vcf.idx $RESDIR/snp/dbSNP/ |
|---|
| 108 | mv dbsnp_138.b37.excluding_sites_after_129.vcf $RESDIR/snp/dbSNP/ |
|---|
| 109 | mv dbsnp_138.b37.excluding_sites_after_129.vcf.idx $RESDIR/snp/dbSNP/ |
|---|
| 110 | mv 1000G_phase1.indels.b37.vcf $RESDIR/sv/1000G |
|---|
| 111 | mv 1000G_phase1.indels.b37.vcf.idx $RESDIR/sv/1000G |
|---|
| 112 | mv testKit1_baits_b37_human_g1k_v37.bed $RESDIR/intervals/ |
|---|
| 113 | mv testKit1_baits_b37_human_g1k_v37.interval_list $RESDIR/intervals/ |
|---|
| 114 | mv testKit1_exons_b37_human_g1k_v37.bed $RESDIR/intervals/ |
|---|
| 115 | mv testKit1_exons_b37_human_g1k_v37.interval_list $RESDIR/intervals/ |
|---|
| 116 | mv 1000G_phase1.indels_Mills_and_1000G_gold_standard.indels.b37.human_g1k_v37.chr1.intervals $RESDIR/intervals/ |
|---|
| 117 | |
|---|
| 118 | ###Remove temporary folder |
|---|
| 119 | rm -r $TMPDIR |
|---|