scripts/make_hg18.sh

   1 #!/bin/sh
   2
   3 #
   4 # Downloads sequence for the hg18 version of H. spiens (human) from
   5 # UCSC.
   6 #
   7 # Note that UCSC's hg18 build has three categories of compressed fasta
   8 # files:
   9 #
  10 # 1. The base files, named chr??.fa.gz
  11 # 2. The unplaced-sequence files, named chr??_random.fa.gz
  12 # 3. The alternative-haplotype files, named chr??_???_hap?.fa.gz
  13 #
  14 # By default, this script builds and index for just the base files,
  15 # since alignments to those sequences are the most useful.  To change
  16 # which categories are built by this script, edit the CHRS_TO_INDEX
  17 # variable below.
  18 #
  19
  20 BASE_CHRS="\
  21 chr1 \
  22 chr2 \
  23 chr3 \
  24 chr4 \
  25 chr5 \
  26 chr6 \
  27 chr7 \
  28 chr8 \
  29 chr9 \
  30 chr10 \
  31 chr11 \
  32 chr12 \
  33 chr13 \
  34 chr14 \
  35 chr15 \
  36 chr16 \
  37 chr17 \
  38 chr18 \
  39 chr19 \
  40 chr20 \
  41 chr21 \
  42 chr22 \
  43 chrX \
  44 chrY \
  45 chrM"
  46
  47 RANDOM_CHRS="\
  48 chr1_random \
  49 chr2_random \
  50 chr3_random \
  51 chr4_random \
  52 chr5_random \
  53 chr6_random \
  54 chr7_random \
  55 chr8_random \
  56 chr9_random \
  57 chr10_random \
  58 chr11_random \
  59 chr13_random \
  60 chr15_random \
  61 chr16_random \
  62 chr17_random \
  63 chr18_random \
  64 chr19_random \
  65 chr21_random \
  66 chr22_random \
  67 chrX_random"
  68
  69 ALT_HAP_CHRS="\
  70 chr22_h2_hap1 \
  71 chr5_h2_hap1 \
  72 chr6_cox_hap1 \
  73 chr6_qbl_hap2"
  74
  75 CHRS_TO_INDEX=$BASE_CHRS
  76
  77 UCSC_HG18_BASE=ftp://hgdownload.cse.ucsc.edu/goldenPath/hg18/chromosomes
  78
  79 get() {
  80         file=$1
  81         if ! wget --version >/dev/null 2>/dev/null ; then
  82                 if ! curl --version >/dev/null 2>/dev/null ; then
  83                         echo "Please install wget or curl somewhere in your PATH"
  84                         exit 1
  85                 fi
  86                 curl -o `basename $1` $1
  87                 return $?
  88         else
  89                 wget $1
  90                 return $?
  91         fi
  92 }
  93
  94 BOWTIE_BUILD_EXE=./bowtie-build
  95 if [ ! -x "$BOWTIE_BUILD_EXE" ] ; then
  96         if ! which bowtie-build ; then
  97                 echo "Could not find bowtie-build in current directory or in PATH"
  98                 exit 1
  99         else
 100                 BOWTIE_BUILD_EXE=`which bowtie-build`
 101         fi
 102 fi
 103
 104 INPUTS=
 105 for c in $CHRS_TO_INDEX ; do
 106         if [ ! -f ${c}.fa ] ; then
 107                 F=${c}.fa.gz
 108                 get ${UCSC_HG18_BASE}/$F || (echo "Error getting $F" && exit 1)
 109                 gunzip $F || (echo "Error unzipping $F" && exit 1)
 110         fi
 111         [ -n "$INPUTS" ] && INPUTS=$INPUTS,${c}.fa
 112         [ -z "$INPUTS" ] && INPUTS=${c}.fa
 113 done
 114
 115 CMD="${BOWTIE_BUILD_EXE} $* ${INPUTS} hg18"
 116 echo Running $CMD
 117 if $CMD ; then
 118         echo "hg18 index built; you may remove fasta files"
 119 else
 120         echo "Index building failed; see error message"
 121 fi