Bio::DB::Fasta -- Fast indexed access to a directory of fasta files
use Bio::DB::Fasta; # create database from directory of fasta files my $db = Bio::DB::Fasta->new('/path/to/fasta/files'); # simple access (for those without Bioperl) my $seq = $db->seq('CHROMOSOME_I',4_000_000 => 4_100_000); my $revseq = $db->seq('CHROMOSOME_I',4_100_000 => 4_000_000); my @ids = $db->ids; my $length = $db->length('CHROMOSOME_I'); my $alphabet = $db->alphabet('CHROMOSOME_I'); my $header = $db->header('CHROMOSOME_I'); # Bioperl-style access my $db = Bio::DB::Fasta->new('/path/to/fasta/files'); my $obj = $db->get_Seq_by_id('CHROMOSOME_I'); my $seq = $obj->seq; my $subseq = $obj->subseq(4_000_000 => 4_100_000); my $length = $obj->length; # (etc) # Bio::SeqIO-style access my $stream = Bio::DB::Fasta->new('/path/to/files')->get_PrimarySeq_stream; while (my $seq = $stream->next_seq) { # Bio::PrimarySeqI stuff } my $fh = Bio::DB::Fasta->newFh('/path/to/fasta/files'); while (my $seq = <$fh>) { # Bio::PrimarySeqI stuff } # tied hash access tie %sequences,'Bio::DB::Fasta','/path/to/fasta/files'; print $sequences{'CHROMOSOME_I:1,20000'};
Bio::DB::Fasta provides indexed access to one or more Fasta files. It provides random access to each sequence entry, and to subsequences within each entry, allowing you to retrieve portions of very large sequences without bringing the entire sequence into memory.
When you initialize the module, you point it at a single fasta file or a directory of multiple such files. The first time it is run, the module generates an index of the contents of the file or directory using the AnyDBM module (Berkeley DB* preferred, followed by GDBM_File, NDBM_File, and SDBM_File). Thereafter it uses the index file to find the file and offset for any requested sequence. If one of the source fasta files is updated, the module reindexes just that one file. (You can also force reindexing manually). For improved performance, the module keeps a cache of open filehandles, closing less-recently used ones when the cache is full.
The fasta files may contain any combination of nucleotide and protein sequences; during indexing the module guesses the molecular type. Entries may have any line length up to 65,536 characters, and different line lengths are allowed in the same file. However, within a sequence entry, all lines must be the same length except for the last.
The module uses /^>(\S+)/ to extract the primary ID of each sequence from the Fasta header. During indexing, you may pass a callback routine to modify this primary ID. For example, you may wish to extract a portion of the gi|gb|abc|xyz nonsense that GenBank Fasta files use. The original header line can be recovered later.
This module was developed for use with the C. elegans and human genomes, and has been tested with sequence segments as large as 20 megabases. Indexing the C. elegans genome (100 megabases of genomic sequence plus 100,000 ESTs) takes ~5 minutes on my 300 MHz pentium laptop. On the same system, average access time for any 200-mer within the C. elegans genome was <0.02s.
*Berkeley DB can be obtained free from www.sleepycat.com. After it is installed you will need to install the BerkeleyDB Perl module.
The two constructors for this class are new() and newFh(). The former creates a Bio::DB::Fasta object which is accessed via method calls. The latter creates a tied filehandle which can be used Bio::SeqIO style to fetch sequence objects in a stream fashion. There is also a tied hash interface.
Create a new Bio::DB::Fasta object from the Fasta file or files indicated by $fasta_path. Indexing will be performed automatically if needed. If successful, new() will return the database accessor object. Otherwise it will return undef.
$fasta_path may be an individual Fasta file, or may refer to a directory containing one or more of such files. Following the path, you may pass a series of name=>value options or a hash with these same name=>value pairs. Valid options are:
Option Name Description Default ----------- ----------- ------- -glob Glob expression to use *.{fa,fasta,fast,FA,FASTA,FAST,dna} for searching for Fasta files in directories. -makeid A code subroutine for None transforming Fasta IDs. -maxopen Maximum size of 32 filehandle cache. -debug Turn on status 0 messages. -reindex Force the index to be 0 rebuilt. -dbmargs Additional arguments none to pass to the DBM routines when tied (scalar or array ref).
-dbmargs can be used to control the format of the index. For example, you can pass $DB_BTREE to this argument so as to force the IDs to be sorted and retrieved alphabetically. Note that you must use the same arguments every time you open the index!
-reindex can be used to force the index to be recreated from scratch.
Create a tied filehandle opened on a Bio::DB::Fasta object. Reading from this filehandle with <> will return a stream of sequence objects, Bio::SeqIO style.
The -makeid option gives you a chance to modify sequence IDs during indexing. The option value should be a code reference that will take a scalar argument and return a scalar result, like this:
$db = Bio::DB::Fasta->new("file.fa",-makeid=>\&make_my_id); sub make_my_id { my $description_line = shift; # get a different id from the fasta header, e.g. $description_line =~ /(\S+)$/; return $1; }
make_my_id() will be called with the full fasta id line (including the ">" symbol!). For example:
>A12345.3 Predicted C. elegans protein egl-2
By default, this module will use the regular expression /^>(\S+)/ to extract "A12345.3" for use as the ID. If you pass a -makeid callback, you can extract any portion of this, such as the "egl-2" symbol.
The -makeid option is ignored after the index is constructed.
The following object methods are provided.
Return the raw sequence (a string) given an ID and optionally a start and stop position in the sequence. In the case of DNA sequence, if $stop is less than $start, then the reverse complement of the sequence is returned (this violates Bio::Seq conventions).
For your convenience, subsequences can be indicated with any of the following compound IDs:
$db->seq("$id:$start,$stop") $db->seq("$id:$start..$stop") $db->seq("$id:$start-$stop")
Return the length of the indicated sequence.
Return the header line for the ID, including the initial ">".
Return the molecular type of the indicated sequence. One of "dna", "rna" or "protein".
Return the name of the file in which the indicated sequence can be found.
Return the offset of the indicated sequence from the beginning of the file in which it is located. The offset points to the beginning of the sequence, not the beginning of the header line.
Return the length of the header line for the indicated sequence.
Return the offset of the header line for the indicated sequence from the beginning of the file in which it is located.
Return the path to the index file.
Return the path to the Fasta file(s).
For BioPerl-style access, the following methods are provided:
Return a Bio::PrimarySeq::Fasta object, which obeys the Bio::PrimarySeqI conventions. For example, to recover the raw DNA or protein sequence, call $seq->seq().
Note that get_Seq_by_id() does not bring the entire sequence into memory until requested. Internally, the returned object uses the accessor to generate subsequences as needed.
These methods all do the same thing as get_Seq_by_id().
Return a Bio::DB::Fasta::Stream object, which supports a single method next_seq(). Each call to next_seq() returns a new Bio::PrimarySeq::Fasta object, until no more sequences remain.
See Bio::PrimarySeqI for methods provided by the sequence objects returned from get_Seq_by_id() and get_PrimarySeq_stream().
This module provides two tied interfaces, one which allows you to treat the sequence database as a hash, and the other which allows you to treat the database as an I/O stream.
The tied hash interface is very straightforward
Tie %db to Bio::DB::Fasta using the indicated path to the Fasta files. The optional @args list is the same set of named argument/value pairs used by Bio::DB::Fasta->new().
If successful, tie() will return the tied object. Otherwise it will return undef.
Once tied, you can use the hash to retrieve an individual sequence by its ID, like this:
my $seq = $db{CHROMOSOME_I};
You may select a subsequence by appending the comma-separated range to the sequence ID in the format "$id:$start,$stop". For example, here is the first 1000 bp of the sequence with the ID "CHROMOSOME_I":
my $seq = $db{'CHROMOSOME_I:1,1000'};
(The regular expression used to parse this format allows sequence IDs to contain colons.)
When selecting subsequences, if $start > stop, then the reverse complement will be returned for DNA sequences.
The keys() and values() functions will return the sequence IDs and their sequences, respectively. In addition, each() can be used to iterate over the entire data set:
while (my ($id,$sequence) = each %db) { print "$id => $sequence\n"; }
When dealing with very large sequences, you can avoid bringing them into memory by calling each() in a scalar context. This returns the key only. You can then use tied(%db) to recover the Bio::DB::Fasta object and call its methods.
while (my $id = each %db) { print "$id => $db{$sequence:1,100}\n"; print "$id => ",tied(%db)->length($id),"\n"; }
You may, in addition invoke Bio::DB::Fasta the FIRSTKEY and NEXTKEY tied hash methods directly.
Return the first ID in the database.
Given an ID, return the next ID in sequence.
This allows you to write the following iterative loop using just the object-oriented interface:
my $db = Bio::DB::Fasta->new('/path/to/fasta/files'); for (my $id=$db->FIRSTKEY; $id; $id=$db->NEXTKEY($id)) { # do something with sequence }
The Bio::DB::Fasta->newFh() method creates a tied filehandle from which you can read Bio::PrimarySeq::Fasta sequence objects sequentially. The following bit of code will iterate sequentially over all sequences in the database:
my $fh = Bio::DB::Fasta->newFh('/path/to/fasta/files'); while (my $seq = <$fh>) { print $seq->id,' => ',$seq->length,"\n"; }
When no more sequences remain to be retrieved, the stream will return undef.
When a sequence is deleted from one of the Fasta files, this deletion is not detected by the module and removed from the index. As a result, a "ghost" entry will remain in the index and will return garbage results if accessed.
Currently, the only way to accomodate deletions is to rebuild the entire index, either by deleting it manually, or by passing -reindex=>1 to new() when initializing the module.
bioperl
Lincoln Stein <lstein@cshl.org>.
Copyright (c) 2001 Cold Spring Harbor Laboratory.
This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself. See DISCLAIMER.txt for disclaimers of warranty.
Title : new Usage : my $db = new Bio::DB::Fasta( $path, @options); Function: initialize a new Bio::DB::Fasta object Returns : new Bio::DB::Fasta object Args : path to dir of fasta files or a single filename
These are optional arguments to pass in as well.
-glob Glob expression to use *.{fa,fasta,fast,FA,FASTA,FAST} for searching for Fasta files in directories. -makeid A code subroutine for None transforming Fasta IDs. -maxopen Maximum size of 32 filehandle cache. -debug Turn on status 0 messages. -reindex Force the index to be 0 rebuilt. -dbmargs Additional arguments none to pass to the DBM routines when tied (scalar or array ref).
Title : newFh Function: gets a new Fh for a file Example : internal method Returns : GLOB Args :
Title : index_dir Usage : $db->index_dir($dir) Function: set the index dir and load all files in the dir Returns : hashref of seq offsets in each file Args : dirname, boolean to force a reload of all files
Title : get_Seq_by_id Usage : my $seq = $db->get_Seq_by_id($id) Function: Bio::DB::RandomAccessI method implemented Returns : Bio::PrimarySeqI object Args : id
Title : set_pack_method Usage : $db->set_pack_method( @files ) Function: Determines whether data packing uses 32 or 64 bit integers Returns : Args : one or more file paths
Title : index_file Usage : $db->index_file($filename) Function: (re)loads a sequence file and indexes sequences offsets in the file Returns : seq offsets in the file Args : filename, boolean to force reloading a file
Title : dbmargs Usage : my @args = $db->dbmargs; Function: gets stored dbm arguments Returns : array Args : none
Title : index_name Usage : my $indexname = $db->index_name($path,$isdir); Function: returns the name of the index for a specific path Returns : string Args : path to check, boolean if it is a dir
Title : calculate_offsets Usage : $db->calculate_offsets($filename,$offsets); Function: calculates the sequence offsets in a file based on id Returns : offset hash for each file Args : file to process $offsets - hashref of id to offset storage
Title : get_all_ids Usage : my @ids = $db->get_all_ids Function: gets all the stored ids in all indexes Returns : list of ids Args : none
Title : subseq Usage : $seqdb->subseq($id,$start,$stop); Function: returns a subseq of a sequence in the db Returns : subsequence data Args : id of sequence, starting point, ending point
Title : get_PrimarySeq_stream Usage : Function: Example : Returns : Args :
To install Bio::Seq, copy and paste the appropriate command in to your terminal.
cpanm
cpanm Bio::Seq
CPAN shell
perl -MCPAN -e shell install Bio::Seq
For more information on module installation, please visit the detailed CPAN module installation guide.