56 lines
1.3 KiB
Bash
56 lines
1.3 KiB
Bash
|
#!/bin/sh
|
||
|
|
|
||
|
|
#
|
||
|
|
# Downloads sequence for the GRCm38 release 81 version of M. Musculus (mouse) from
|
||
|
|
# Ensembl.
|
||
|
|
#
|
||
|
|
# By default, this script builds and index for just the base files,
|
||
|
|
# since alignments to those sequences are the most useful. To change
|
||
|
|
# which categories are built by this script, edit the CHRS_TO_INDEX
|
||
|
|
# variable below.
|
||
|
|
#
|
||
|
|
|
||
|
|
ENSEMBL_RELEASE=84
|
||
|
|
ENSEMBL_GRCm38_BASE=ftp://ftp.ensembl.org/pub/release-${ENSEMBL_RELEASE}/fasta/mus_musculus/dna
|
||
|
|
|
||
|
|
get() {
|
||
|
|
file=$1
|
||
|
|
if ! wget --version >/dev/null 2>/dev/null ; then
|
||
|
|
if ! curl --version >/dev/null 2>/dev/null ; then
|
||
|
|
echo "Please install wget or curl somewhere in your PATH"
|
||
|
|
exit 1
|
||
|
|
fi
|
||
|
|
curl -o `basename $1` $1
|
||
|
|
return $?
|
||
|
|
else
|
||
|
|
wget $1
|
||
|
|
return $?
|
||
|
|
fi
|
||
|
|
}
|
||
|
|
|
||
|
|
HISAT2_BUILD_EXE=./hisat2-build
|
||
|
|
if [ ! -x "$HISAT2_BUILD_EXE" ] ; then
|
||
|
|
if ! which hisat2-build ; then
|
||
|
|
echo "Could not find hisat2-build in current directory or in PATH"
|
||
|
|
exit 1
|
||
|
|
else
|
||
|
|
HISAT2_BUILD_EXE=`which hisat2-build`
|
||
|
|
fi
|
||
|
|
fi
|
||
|
|
|
||
|
|
rm -f genome.fa
|
||
|
|
F=Mus_musculus.GRCm38.dna.primary_assembly.fa
|
||
|
|
if [ ! -f $F ] ; then
|
||
|
|
get ${ENSEMBL_GRCm38_BASE}/$F.gz || (echo "Error getting $F" && exit 1)
|
||
|
|
gunzip $F.gz || (echo "Error unzipping $F" && exit 1)
|
||
|
|
mv $F genome.fa
|
||
|
|
fi
|
||
|
|
|
||
|
|
CMD="${HISAT2_BUILD_EXE} genome.fa genome"
|
||
|
|
echo Running $CMD
|
||
|
|
if $CMD ; then
|
||
|
|
echo "genome index built; you may remove fasta files"
|
||
|
|
else
|
||
|
|
echo "Index building failed; see error message"
|
||
|
|
fi
|