High Performance Computing

Versions and Availability

About the Software

Basic Local Alignment Search Tool, or BLAST, is an algorithm for comparing primary biological sequence information, such as the amino-acid sequences of different proteins or the nucleotides of DNA sequences. - Homepage: http://blast.ncbi.nlm.nih.gov/

Usage

A suite of tools are provided in the BLAST+, such as blastx, blastn, blastp. The command line below is provided by NCBI, which is a blastn search against a database. This command:

    blastn -db nt -qurey nt.fsa -out results.out
    

will run a blastn search of nt.fsa (a nucleotide sequence in FASTA format) against the nt database, printing results to the file results.out.

BLAST+ uses an environment variable $BLASTDB to point to the directory where the database is located in. So this environment variable should be specified before running blast commands, see example below:

Example: run blastx search in PBS batch job

This example shows a blastx search of trinity.fasta against the nr database, printing results to the file blastx.out. The nr database files are in the directory /work/ychen64/nr.

#!/bin/bash
#PBS -q workq
#PBS -l nodes=1:ppn=20
#PBS -l walltime=72:00:00
#PBS -A your_allocation_name
export BLASTDB=/work/ychen64/nr
blastx -query trinity.fasta -db nr -out blastx.out -num_threads 20
    

The option -num_threads is used to specify the number of threads (CPUs) in the BLAST search. This number should match the number in "ppn=" in the "#PBS -l nodes=1:ppn=" to fully utilize the CPU power.

Example: create local BLAST database in PBS batch job

The local BLAST database can be created by a perl script called "update_blastdb.pl" included as part of BLAST+. perl is required to run this script. In this example, a nr database is created at /work/ychen64

#!/bin/bash
#PBS -q workq
#PBS -l nodes=1:ppn=20
#PBS -l walltime=72:00:00
#PBS -A your_allocation_name
cd /work/ychen64
# Specify the database name here:
export DATABASE=nr

mkdir $DATABASE
cd $DATABASE
update_blastdb.pl --decompress --verbose $DATABASE
    

It will take a while to create a large local BLAST database, so update_blastdb.pl should be run with the interactive or batch job.

Once the local BLAST database is created, it needs to be updated in a timely manner (every couple of days/weeks months based on your database). Just run the script above again to update the database. The Documentation for the update_blastdb.pl script is available by running the script without any arguments.

Note:

• Please don't search against the database on the NCBI BLAST server (i.e. using "-remote" option), especially for the large case. Search against local BLAST database only.
• Please only use one compute node to run BLAST job. The parallel search technique provided by BLAST+ is based on the thread parallelism , so it cannot be used on the multiple nodes.
• On the cluster with SLURM job scheduler (rather than PBS), use #SBATCH -c to specify the number of threads per process in the SLURM directives. Skip #SBATCH -n in the SLURM directives.

Resources

• The BLAST Home Page provides links to protein and genomic data sets, as well as information on specific tools.