Using BioPerl to create, manage, and query BLAST databases using the NCBI’s blast+
suite.
Quicklink to Synopsis.
Mark A. Jensen
maj -at- fortinbras -dot- us
blast+ is a suite of programs from NCBI that creates, manipulates, manages and queries BLAST sequence databases. The package attempts to integrate the many blast+
programs into a wrapper providing a unified programmatic interface to these programs, with BioPerl objects as input and output if desired.
The module and dependencies are available in the bioperl/bioperl-run git repository. To download and install BioPerl core and packages from github see INSTALL.
Like all run wrappers, these modules need the underlying programs to work. Get blast+ 2.2.22
at the NCBI FTP site. The user manual is very helpful.
The Bio::Tools::Run::StandAloneBlastPlus object is a “factory” or harness that directs the execution of the various blast+
programs. The basic mantra is to (1) create a StandAloneBlastPlus
factory using the new()
constructor, and (2) perform BLAST analyses by calling the desired BLAST program by name off the factory object. The database can be pre-existing, or can be created directly using a FASTA file or a BioPerl sequence collection object. Low-complexity or other masking can also be applied as the database is constructed.
The BLAST database itself and any masking data are attached to the factory object (step 1). Query sequences and any parameters associated with particular programs are provided to the blast method call (step 2), and are run against the attached database. (We present step 2 first, since it’s what people will do many times after creating their database once.)
blast+
also provides facilities for blasting sequences against NCBI databases over the network. See Remote BLAST for details.
Given a StandAloneBlastPlus
factory, such as
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'testdb'
);
(see factory construction below) you can run the desired BLAST method directly from the factory object, against the database currently attached to the factory (in the above example, testdb
). -query
is a required argument:
$result = $fac->blastn( -query => 'query_seqs.fas' );
Here, $result
is a Bio::Search::Result::BlastResult object. To obtain further results, use next_result
:
while ($result = $fac->next_result) {
...
}
Rewind to the beginning of the results:
$fac->rewind_results;
while ($result = $fac->next_result) {
...
}
BLAST methods include blastn, blastp, blastx, tblastn, tblastx, rpsblast, psiblast, rpstblastn
, as well as others I probably forget. For more details on parsing and processing the results, see the SearchIO HOWTO.
Note the following details:
$result = $fac->blastn( -query => 'query_seqs.fas',
-outfile => 'query.bls');
$result = $fac->blastn( -query => 'query_seqs.fas',
-outfile => 'query.bls',
-outformat => 5); # 5=XML output
$result = $fac->blastn( -query => 'query_seqs.fas',
-outfile => 'query.bls',
-method_args => [ -num_alignments => 10,
-evalue => 100 ]);
$file = $fac->blast_out;
$fac->cleanup;
Running bl2seq
is similar to running the methods as outlined above, but both -query
and -subject
parameters are required required, and the attached database is ignored. The BLAST method must be specified explicitly with the -method
parameter:
$fac->bl2seq( -method => 'blastp',
-query => $seq_object_1,
-subject => $seq_object_2);
Other parameters ( -method_args
, -outfile
, and -outformat
) are valid.
The return value is always a Bio::Search::Result::BlastResult object on success, undef
on failure.
First of all, the factory needs to be told where the blast+
programs live. The BLASTPLUSDIR
environment variable will be checked for the default executable directory. Alternatively, the program directory can be set for individual factory instances using the -prog_dir
constructor parameter. All the blast+
programs must be accessible from that directory (i.e., as executable files or symlinks).
Either the database or BLAST subject data must be specified at factory construction. Databases can be pre-existing formatted BLAST dbs, or can be built directly from fasta sequence files or BioPerl sequence object collections of several kinds. The key constructor parameters are -db_name
, -db_data
, and -db_dir
.
To specify a pre-existing BLAST database, use -db_name
alone:
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'mydb'
);
The directory can be specified along with the basename, or separately with -db_dir
:
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => '~/home/blast/mydb'
);
# same as
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'mydb', -db_dir => '~/home/blast'
);
To create a BLAST database de novo, see Creating a BLAST database.
If you wish to apply pre-existing mask data (i.e., the final ASN1 output from one of the blast+
masker programs), to the database before querying, specify it with -mask_file
:
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'mydb', -mask_file => 'mymaskdata.asn'
);
There are several options for creating the database de novo using attached data, both before and after factory construction. If a temporary database (one that can be deleted by the cleanup()
method) is desired, leave out the -db_name
parameter. If -db_name
is specified, the database will be preserved with the basename specified.
Use -create => 1
to create a new database (otherwise the factory will look for an existing database). Use -overwrite => 1
to create and overwrite an existing database.
Note that the database is not created immediately on factory construction. It will be created if necessary on the first use of a factory BLAST method, or you can force database creation by executing
$fac->make_db();
With a FASTA file:
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'mydb',
-db_data => 'myseqs.fas',
-create => 1
);
With another BioPerl object collection:
$alnio = Bio::AlignIO->new( -file => 'alignment.msf' );
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'mydb',
-db_data => $alnio,
-create => 1
);
@seqs = $alnio->next_aln->each_seq;
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'mydb',
-db_data => @seqs,
-create => 1
);
Other collections (e.g., Bio::SeqIO) are valid. If a certain type does not work, please open a new issue.
To create temporary databases, leave out the -db_name
, e.g.
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_data => 'myseqs.fas',
-create => 1
);
To get the tempfile basename, do:
$dbname = $fac->db;
After the factory has been created, you can use the explict attribute setters:
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-create => 1
);
$fac->set_db_data('myseqs.fas');
$fac->make_db;
and so on.
The blast+ mask utilities windowmasker
, segmasker
, and dustmasker
are available. Masking can be rolled into database creation, or can be executed later. If your mask data is already created and in ASN1 format, set the -mask_file
attribute on construction (see Factory construction/initialization).
To create a mask from raw data or an existing database and apply the mask upon database creation, construct the factory like so:
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'my_masked_db',
-db_data => 'myseqs.fas',
-masker => 'dustmasker',
-mask_data => 'maskseqs.fas',
-create => 1
);
The masked database will be created during make_db()
.
The -mask_data
parameter can be a FASTA filename or any BioPerl sequence object collection. If the datatype (nucl
or prot
) of the mask data is not compatible with the selected masker, an exception will be thrown with a message to that effect.
To create a mask ASN1 file that can be used in the -mask_file
parameter separately from the attached database, use the make_mask()
method directly:
$mask_file = $fac->make_mask(-data => 'maskseqs.fas',
-masker => 'dustmasker');
# segmasker can use a blastdb as input
$mask_file = $fac->make_mask(-mask_db => 'mydb',
-masker => 'segmasker')
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'my_masked_db',
-db_data => 'myseqs.fas',
-mask_file => $mask_file
-create => 1
);
To get a hash containing useful metadata on an existing database (obtained by running blastdbcmd -info
), use db_info()
:
# get info on the attached database..
$info = $fac->db_info;
# get info on another database
$info = $fac->db_info('~/home/blastdbs/another');
To get a particular info element for the attached database, just call the element name off the factory:
$num_seqs = $fac->db_num_sequences;
# info on all the masks applied to the db, if any:
@masking_info = @{ $fac->db_filter_algorithms };
The blast+
programs are actually executed by a Bio::Tools::Run::BlastPlus wrapper instance. This instance is available for peeking and poking in the StandAloneBlastPlus
factory()
attribute. For convenience, BlastPlus
methods can be run directly from the StandAloneBlastPlus
object, and are delegated to the factory()
attribute.
For example, to get the blast+ program to be executed, examine either
$fac->factory->command
or
$fac->command
Similarly, the currently set parameters for the BlastPlus
factory are
@parameters = $fac->get_parameters
To get a list of what parameters you can set for the current blast+
operation, do:
@available = $fac->available_parameters('all');
Temporary analysis files produced under a single factory instances can be unlinked by running
$fac->cleanup;
Tempfiles are generally not removed unless this method is explicitly called. Note that cleanup()
only unlinks “registered” files and databases. All temporary files are automatically registered; in particular, “anonymous” databases (such as
$fac->Bio::Tools::Run::StandAloneBlastPlus->new(
-db_data => 'myseqs.fas',
-create => 1
);
without a -db_name
specification) are registered for cleanup. Any file or database can be registered with an internal method:
$fac->_register_temp_for_cleanup('testdb');
StandAloneBlastPlus
can access NCBI databases remotely. Just create a factory with the desired database and -remote => 1
:
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'nr',
-remote => 1
);
and call BLAST methods from the factory as usual. Note that the database info methods are unavailable for remote databases. Available remote databases are listed here.
$is_good = $fac->check_db('putative_db');
blast+
programs (if you know what you’re doing) with db_make_args
and mask_make_args
:$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'customdb',
-db_data => 'myseqs.fas',
-db_make_args => [ -taxid_map => 'seq_to_taxa.txt' ],
-masker => 'windowmasker',
-mask_data => 'myseqs.fas',
-mask_make_args => [ -dust => 'T' ],
-create => 1
);
blast+
program executions by setting no_throw_on_crash()
. Examine the error with stderr()
:$fac->no_throw_on_crash(1);
$fac->make_db;
if ($fac->stderr =~ /Error:/) {
# handle error
...
}
# existing blastdb:
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'mydb'
);
# create blastdb from fasta file and attach
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'mydb',
-db_data => 'myseqs.fas',
-create => 1
);
# create blastdb from BioPerl sequence collection objects
$alnio = Bio::AlignIO->new( -file => 'alignment.msf' );
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'mydb',
-db_data => $alnio,
-create => 1
);
# blast against the remote NCBI db
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'nr',
-remote => 1
);
@seqs = $alnio->next_aln->each_seq;
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'mydb',
-db_data => @seqs,
-create => 1
);
# create database with masks
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'my_masked_db',
-db_data => 'myseqs.fas',
-masker => 'dustmasker',
-mask_data => 'maskseqs.fas',
-create => 1
);
# create a mask datafile separately
$mask_file = $fac->make_mask(
-data => 'maskseqs.fas',
-masker => 'dustmasker'
);
# query database for metadata
$info_hash = $fac->db_info;
$num_seq = $fac->db_num_sequences;
@mask_metadata = @{ $fac->db_filter_algorithms };
# perform blast methods
$result = $fac->tblastn( -query => $seqio );
# create a factory:
$fac = Bio::Tools::Run::StandAloneBlastPlus->new(
-db_name => 'testdb'
);
# get your results
$result = $fac->blastn(-query => 'query_seqs.fas',
-outfile => 'query.bls',
-method_args => [ -num_alignments => 10 ] );
$result = $fac->tblastx(-query => $an_alignment_object,
-outfile => 'query.bls',
-outformat => 7 );
# do a bl2seq
$fac->bl2seq(-method => 'blastp',
-query => $seq_object_1,
-subject => $seq_object_2 );