| 1 |
#!/bin/sh |
| 2 |
# Script to update a local uniprot copy with sequence files and generate blast dbs for them (formatdb) |
| 3 |
|
| 4 |
if [ -z "$1" ] |
| 5 |
then |
| 6 |
echo "Usage: $0 <base local dir>" |
| 7 |
exit 1 |
| 8 |
fi |
| 9 |
|
| 10 |
|
| 11 |
LOCALDIR=$1 |
| 12 |
CURRENT="$LOCALDIR/current" |
| 13 |
DOWNLOAD="$LOCALDIR/download" |
| 14 |
|
| 15 |
#SITE="ftp://ftp.uniprot.org/pub" # US main ftp |
| 16 |
SITE="ftp://ftp.ebi.ac.uk/pub" # UK mirror |
| 17 |
# the swiss mirror doesn't seem to update properly, not using it anymore |
| 18 |
#SITE="ftp://ftp.expasy.org" # swiss mirror |
| 19 |
|
| 20 |
|
| 21 |
COMPLETEKBDIR="databases/uniprot/current_release/knowledgebase/complete" |
| 22 |
|
| 23 |
SIFTSPDB2UNIPROTFTP="ftp://ftp.ebi.ac.uk/pub/databases/msd/sifts/text/pdb_chain_uniprot.lst" |
| 24 |
|
| 25 |
|
| 26 |
SPROT="uniprot_sprot.fasta" |
| 27 |
SPROTGZ="${SPROT}.gz" |
| 28 |
TREMBL="uniprot_trembl.fasta" |
| 29 |
TREMBLGZ="${TREMBL}.gz" |
| 30 |
ALL="uniprot_all.fasta" |
| 31 |
RELDATEFILE="reldate.txt" |
| 32 |
SIFTSPDB2UNIPROT="pdb_chain_uniprot.lst" |
| 33 |
|
| 34 |
sproturl="$SITE/$COMPLETEKBDIR/$SPROTGZ" |
| 35 |
tremblurl="$SITE/$COMPLETEKBDIR/$TREMBLGZ" |
| 36 |
reldateurl="$SITE/$COMPLETEKBDIR/$RELDATEFILE" |
| 37 |
|
| 38 |
# remove existing download directory if there was one |
| 39 |
rm -rf $DOWNLOAD |
| 40 |
# create the download dir |
| 41 |
mkdir $DOWNLOAD |
| 42 |
|
| 43 |
# getting the release date file if newer available |
| 44 |
release="" |
| 45 |
curl -z $CURRENT/$RELDATEFILE $reldateurl > $DOWNLOAD/$RELDATEFILE |
| 46 |
if [ -s "$DOWNLOAD/$RELDATEFILE" ] |
| 47 |
then |
| 48 |
release=`head -1 $DOWNLOAD/$RELDATEFILE | sed "s/UniProt Knowledgebase Release \(...._..\).*/\1/"` |
| 49 |
echo "New uniprot release $release available. Downloading files." |
| 50 |
else |
| 51 |
echo "No new uniprot release available. Exiting" |
| 52 |
rm -rf $DOWNLOAD |
| 53 |
exit 0 |
| 54 |
fi |
| 55 |
|
| 56 |
|
| 57 |
# download if newer available |
| 58 |
curl -z $CURRENT/$TREMBL $tremblurl > $DOWNLOAD/${TREMBL}.gz |
| 59 |
if [ -s "$DOWNLOAD/${TREMBL}.gz" ] |
| 60 |
then |
| 61 |
echo "New trembl version downloaded" |
| 62 |
else |
| 63 |
echo "Remote trembl file not newer than local one. Something wrong. Exiting." |
| 64 |
exit 1 |
| 65 |
fi |
| 66 |
|
| 67 |
curl -z $CURRENT/$SPROT $sproturl > $DOWNLOAD/${SPROT}.gz |
| 68 |
if [ -s "$DOWNLOAD/${SPROT}.gz" ] |
| 69 |
then |
| 70 |
echo "New sprot version downloaded" |
| 71 |
else |
| 72 |
echo "Remote sprot file not newer than local one. Something wrong. Exiting." |
| 73 |
exit 1 |
| 74 |
fi |
| 75 |
|
| 76 |
|
| 77 |
# getting the SIFTS PDB to UNIPROT mapping file |
| 78 |
curl $SIFTSPDB2UNIPROTFTP > $DOWNLOAD/$SIFTSPDB2UNIPROT |
| 79 |
|
| 80 |
|
| 81 |
# uncompressing |
| 82 |
gzip -df $DOWNLOAD/${SPROT}.gz |
| 83 |
gzip -df $DOWNLOAD/${TREMBL}.gz |
| 84 |
# creating the "all" file |
| 85 |
cat $DOWNLOAD/$TREMBL $DOWNLOAD/$SPROT > $DOWNLOAD/$ALL |
| 86 |
|
| 87 |
# run formatdb |
| 88 |
# formatdb appends the path used to run it to the .pal index file, |
| 89 |
# thus if the path used is an absolute path it's effectively hard coding |
| 90 |
# them making the directory not movable. That's why we have to cd to the |
| 91 |
# DOWNLOAD dir first, so that there's no hard-coded paths in the .pal file |
| 92 |
|
| 93 |
|
| 94 |
echo "Running formatdb..." |
| 95 |
|
| 96 |
#formatdb log file |
| 97 |
logfile="$DOWNLOAD/formatdb.log" |
| 98 |
|
| 99 |
cd $DOWNLOAD |
| 100 |
formatdb -p T -o T -l $logfile -i $SPROT |
| 101 |
formatdb -p T -o T -l $logfile -i $TREMBL |
| 102 |
formatdb -p T -o T -l $logfile -i $ALL |
| 103 |
|
| 104 |
#renaming DOWNLOAD dir to uniprot version and updating current symlink |
| 105 |
echo "Creating new symlink..." |
| 106 |
mv $DOWNLOAD $LOCALDIR/uniprot_$release |
| 107 |
rm -f $CURRENT |
| 108 |
cd $LOCALDIR |
| 109 |
ln -s uniprot_$release current |
| 110 |
|
| 111 |
echo "Done" |
