README.blast

# GRAMENE BLAST; How to set up the blast server
#==============================================

# These instructions were tested for the versions below on 
# catskill.cshl.edu 

# CODE INSTALLATION
#------------------

# bash shell assumed
# Set code/data versions (will change with each release)
# These are just used for convenience for this README
export ENSVER=65
export GRAVER=34b
export EGVER=12

# Fetch the version-specific code into a clean directory
mkdir /usr/local/svntmp/ensembl-grmblast-$ENSVER
cd /usr/local/svntmp/ensembl-grmblast-$ENSVER

cvs -d :pserver:cvsuser@cvsro.sanger.ac.uk:/cvsroot/ensembl login
[pass CVSUSER]

cvs -d :pserver:cvsuser@cvsro.sanger.ac.uk:/cvsroot/ensembl \
co -r branch-ensembl-$ENSVER ensembl-api

cvs -d :pserver:cvsuser@cvsro.sanger.ac.uk:/cvsroot/ensembl \
co -r branch-ensembl-$ENSVER ensembl-website

cvs -d :pserver:cvsuser@cvsro.sanger.ac.uk:/cvsroot/ensembl \
co -r branch-ensemblgenomes-$EGVER-$ENSVER eg-plugins

svn co http://svn.warelab.org/gramene/branches/build${GRAVER}-branch \
gramene-$GRAVER

# Move the code directory into place
sudo mv /usr/local/svntmp/ensembl-grmblast-$ENSVER /usr/local

# Create links to code that does not change
cd /usr/local/ensembl-grmblast-$ENSVER 
ln -s gramene-$GRAVER gramene-live
ln -s /usr/local/bioperl-1-2-3 bioperl-live
ln -s /usr/local/biomart-perl-0_6 biomart-perl
ln -s /usr/local/apache2 apache2
cp /usr/local/vcftools_0.1.5/lib/Vcf.pm /usr/local/ensembl-grmblast/modules/

# Create temporary dirs
mkdir logs
mkdir -p /usr/local/ensembl-grmblast/htdocs/BlastView/minified

# Move the Ensembl Plugins file into place
cp gramene-live/ensembl-plugins/grmblast/conf/Plugins.pm conf/

# Stop the old server, and start the new
# NOTE: the new server should have been tested first!

sudo kill `cat /usr/local/ensembl-grmblast/logs/httpd.pid`
sudo rm /usr/local/ensembl-grmblast
sudo ln -s /usr/local/ensembl-grmblast-$ENSVER /usr/local/ensembl-grmblast
sudo /etc/init.d/apachectl-grmblast start


# DATA INSTALLATION
#------------------

# Get the version-specific FASTA databases
# Note that we download these from the EnsemblGenomes FTP site, 
# But they can be generated direct from the Ensembl databases
# using the utils/do_fasta_dump script

##get sequences
# for EG genomes, we can download from EG ftp site

wget -nH -r --cut-dirs=3 \
"ftp://ftp.ensemblgenomes.org/pub/release-$EGVER/plants/fasta/"

# for Gramene specific genomes such as OGE genomes, create locally
cd ensembl-live/
perl  utils/do_fasta_dump  
	--no_ssaha --species $s 
	-type dna_toplevel -type cdna_all -type pep_all -type ncrna 
	-dumpdir OGEdump/ 
	-release 71 
	--no_log 
	-email weix@cshl.edu;

# Index the FASTA DBs
find fasta/ -name \*.gz -exec gunzip {} \;
find fasta/ -name \*.pep.\*.fa -exec /usr/local/wublast/xdformat -p {} \;
find fasta/ -name \*dna\*.fa -exec /usr/local/wublast/xdformat -n {} \;

# Move the indexes into place, and update the symlink
mkdir /usr/local/blastdb/gramene-$ENSVER
find fasta/ -name \*.fa.x?? -exec mv {} /usr/local/blastdb/gramene-$ENSVER \;
sudo rm /usr/local/blastdb/gramene
sudo ln -s /usr/local/blastdb/gramene-$ENSVER \
/usr/local/blastdb/gramene

# Some slight renaming to convert from EG to Ensembl convention
cd /usr/local/blastdb/gramene
ls | perl -e 'while(<>){ chomp; my $ori=$_; my $new=$ori; if( $ori =~ /^.*\..*\.(.*\.).*\..*\.fa/..*$ ){ $new =~ s/$1//; `mv $ori $new` } }'
For example, the filename should be Oryza_sativa.IRGSP-1.0.dna.toplevel.fa.* or Zea_mays.AGPv3.pep.all.fa.*
The format should be, and we only need *.dna.toplevel.fa.* and *.pep.all.fa.*, we do not need the individual chromosome files to be indexed,
Species.assemblyVersion.dna.toplevel.fa.*
Species.assemblyVersion.pep.all.fa.*

# MAINTENANCE
#------------

Ensure that there is a startup script in /etc/init.d that is correctly
linked from /etc/rc.d to ensure that the web server is restarted on
server reboot.

There are also some scripts that need installing in the root user's
crontab that clean up temporary blast files and database records. Here
is an example (please replace <ENSROOT> with the value of the $ENSROOT
variable);

---
# Clean ensembl* tmp and img-tmp files, and the blast database
0  4 * * * find <ENSROOT>/*tmp -type f -atime +1 -exec rm {} \; > /dev/null 2>&1
0  4 * * * find <ENSROOT>/img-cache -type f -atime +1 -exec rm {} \; > /dev/null 2>&1
0  3 * * * perl <ENSROOT>/utils/blast_cleaner.pl 
---