How
to Set Up a BLAST Web Server
Do you
need to provide local Web access‚ with an NCBI-like interface‚
to BLAST‚ PSI-BLAST‚ PHI-BLAST‚ MegaBLAST‚ BLAST2Sequences‚
and RPS-BLAST? Do you need to support batch queries by sequence ID or
accession number? NCBI offers a Web Server package that will support these
interfaces as well as batch searching‚ XML output‚ and more.
Step 1: FTP the Web Server Package
Download
the latest Web Server package by FTP from:
ftp://ncbi.nlm.nih.gov/blast/server/
Files
are located in the “current_release” directory. Releases are
available for Linux/Intel‚ OSF/DEC-Alpha‚ SGI and Solaris/Sun
or Intel.
Step 2: Unpack the Archive
Unpack the distribution into any local document directory under the HTTPD server directory that has permission to run CGI programs with a *.cgi file extension. The BLAST Web server package is a Gzipped‚ tarred archive. Gunzip the archive using a command such as:
gzip
-d wwwblast.platform.tar.gz
or
gunzip wwwblast.platform.tar.gz
Next‚
untar the archive using:
tar -xvpf wwwblast.platform.tar
Note
the use of the “p”‚ or “preserve” TAR option. This option preserves the
file protections as they existed when the files were archived. This is
essential to allow the proper access to all files by the server software.
Uncompressing the archive will create a directory called “blast”‚ which
contains all the files needed to implement the WWW BLAST server.
Step 3: Format Local Databases
Format the local databases that you wish the server to use. This is accomplished in the usual manner using the program “formatdb”‚ included in the server package. The database files should be placed in the “db” directory‚ which is created when you untar the archive. To format a database for BLAST‚ use:
formatdb -i infile -o T
Note
that databases to be used with PSI- or PHI-BLAST should always be created
using the “-o T” flag.
Step
4: Test the Installation
The
server package comes with two BLAST databases called “test_aa_db”—a
sample protein database‚ and “test_na_db”—a sample
nucleotide database. These databases are configured to be used immediately.
Perform a test search on these database using the “blast.html”
Web page.
Step 5: Configure Local Databases
To
add your own databases to the list used by the server‚ you must edit
two configuration files called “blast.rc” and “psiblast.rc”.
These files specify the combinations of database names and BLAST program
that are considered valid. Below is an excerpt from the file “blast.rc”
as it appears “out of the box.”
#
Here is list of program/database‚
# that allowed by BLAST service.
# Format:<program> <db> <db> ...
#
blastn
test_na_db
blastp test_aa_db
blastx test_aa_db
tblastn test_na_db
tblastx test_na_db
The line “blastn test_na_db‚” for example‚ specifies that the database named “test_na_db” is valid when used with the blastn program. To add another nucleotide database to the list searchable by blastn‚ simply add its name to this line so that it now reads:
blastn test_na_db new_database
The
new database name must also be added to the html search forms in order
to appear in the pull-down database menu boxes‚ so the blast.html
file also needs to be edited. The relevant lines for blast.html are:
<select name = “DATALIB”>
<option
VALUE = “test_na_db”>
test_na_db
<option
VALUE = “test_aa_db”>
test_aa_db
<option VALUE=
“test_aa_db
pdb”>
test_aa_db
& pdb
<option
VALUE = “pdbaa”> PDB
</select>
In this case‚ the first two “option” lines specify the
two test databases. The third “option” line specifies that a
database labeled “test_aa_db & pdb” can be selected from the
pull-down list box. If a user selects this database‚ BLAST will search
two databases‚ “test_aa_db” and “pdb‚” as
though they were a single database. The fourth “option” line
indicates that a database called “pdbaa” will be labeled in
the pull-down list box as “PDB.” Other html search forms‚
such as the form for MegaBLAST‚ will also require the same type of
alteration when a new database is added.
Step 6: Celebrate
Run your first flawless test of the BLAST server!
|
