scripts/make_canFam2.sh

   1 #!/bin/sh
   2
   3 #
   4 # Downloads sequence for the canFam2 version of C. familiaris (dog)
   5 # from UCSC.
   6 #
   7
   8 i=2
   9 BASE_CHRS=chr1
  10 while [ $i -lt 39 ] ; do
  11         BASE_CHRS="$BASE_CHRS chr$i"
  12         i=`expr $i + 1`
  13 done
  14 BASE_CHRS="$BASE_CHRS chrX chrM chrUn"
  15 CHRS_TO_INDEX=$BASE_CHRS
  16
  17 CANFAM2_BASE=ftp://hgdownload.cse.ucsc.edu/goldenPath/canFam2/chromosomes
  18
  19 get() {
  20         file=$1
  21         if ! wget --version >/dev/null 2>/dev/null ; then
  22                 if ! curl --version >/dev/null 2>/dev/null ; then
  23                         echo "Please install wget or curl somewhere in your PATH"
  24                         exit 1
  25                 fi
  26                 curl -o `basename $1` $1
  27                 return $?
  28         else
  29                 wget $1
  30                 return $?
  31         fi
  32 }
  33
  34 BOWTIE_BUILD_EXE=./bowtie-build
  35 if [ ! -x "$BOWTIE_BUILD_EXE" ] ; then
  36         if ! which bowtie-build ; then
  37                 echo "Could not find bowtie-build in current directory or in PATH"
  38                 exit 1
  39         else
  40                 BOWTIE_BUILD_EXE=`which bowtie-build`
  41         fi
  42 fi
  43
  44 INPUTS=
  45 for c in $CHRS_TO_INDEX ; do
  46         if [ ! -f ${c}.fa ] ; then
  47                 F=${c}.fa.gz
  48                 get ${CANFAM2_BASE}/$F || (echo "Error getting $F" && exit 1)
  49                 gunzip $F || (echo "Error unzipping $F" && exit 1)
  50         fi
  51         [ -n "$INPUTS" ] && INPUTS=$INPUTS,${c}.fa
  52         [ -z "$INPUTS" ] && INPUTS=${c}.fa
  53 done
  54
  55 CMD="${BOWTIE_BUILD_EXE} $* ${INPUTS} canFam2"
  56 echo Running $CMD
  57 if $CMD ; then
  58         echo "canFam2 index built; you may remove fasta files"
  59 else
  60         echo "Index building failed; see error message"
  61 fi