BioRuby Tutorial

This document was last modified: 2011/10/14 Current editor: Michael O'Keefe <okeefm (at) rpi (dot) edu>

The latest version resides in the GIT source code repository: ./doc/Tutorial.rd.

Introduction

This is a tutorial for using Bioruby. A basic knowledge of Ruby is required. If you want to know more about the programming language, we recommend the latest Ruby book Programming Ruby by Dave Thomas and Andy Hunt - the first edition can be read online here.

For BioRuby you need to install Ruby and the BioRuby package on your computer

You can check whether Ruby is installed on your computer and what version it has with the

% ruby -v

command. You should see something like:

ruby 1.9.2p290 (2011-07-09 revision 32553) [i686-linux]

If you see no such thing you'll have to install Ruby using your installation manager. For more information see the Ruby website.

With Ruby download and install Bioruby using the links on the Bioruby website. The recommended installation is via RubyGems:

gem install bio

See also the Bioruby wiki.

A lot of BioRuby's documentation exists in the source code and unit tests. To really dive in you will need the latest source code tree. The embedded rdoc documentation can be viewed online at bioruby's rdoc. But first lets start!

Trying Bioruby

Bioruby comes with its own shell. After unpacking the sources run one of the following commands:

bioruby

or, from the source tree

cd bioruby
ruby -I lib bin/bioruby

and you should see a prompt

bioruby>

Now test the following:

bioruby> require 'bio'
bioruby> seq = Bio::Sequence::NA.new("atgcatgcaaaa")
==> "atgcatgcaaaa"

bioruby> seq.complement
==> "ttttgcatgcat"

See the the Bioruby shell section below for more tweaking. If you have trouble running examples also check the section below on trouble shooting. You can also post a question to the mailing list. BioRuby developers usually try to help.

Working with nucleic / amino acid sequences (Bio::Sequence class)

The Bio::Sequence class allows the usual sequence transformations and translations. In the example below the DNA sequence "atgcatgcaaaa" is converted into the complemental strand and spliced into a subsequence; next, the nucleic acid composition is calculated and the sequence is translated into the amino acid sequence, the molecular weight calculated, and so on. When translating into amino acid sequences, the frame can be specified and optionally the codon table selected (as defined in codontable.rb).

bioruby> seq = Bio::Sequence::NA.new("atgcatgcaaaa")
==> "atgcatgcaaaa"

# complemental sequence (Bio::Sequence::NA object)
bioruby> seq.complement
==> "ttttgcatgcat"

bioruby> seq.subseq(3,8) # gets subsequence of positions 3 to 8 (starting from 1)
==> "gcatgc"
bioruby> seq.gc_percent 
==> 33
bioruby> seq.composition 
==> {"a"=>6, "c"=>2, "g"=>2, "t"=>2}
bioruby> seq.translate 
==> "MHAK"
bioruby> seq.translate(2)        # translate from frame 2
==> "CMQ"
bioruby> seq.translate(1,11)     # codon table 11
==> "MHAK"
bioruby> seq.translate.codes
==> ["Met", "His", "Ala", "Lys"]
bioruby> seq.translate.names
==> ["methionine", "histidine", "alanine", "lysine"]
bioruby>  seq.translate.composition
==> {"K"=>1, "A"=>1, "M"=>1, "H"=>1}
bioruby> seq.translate.molecular_weight
==> 485.605
bioruby> seq.complement.translate
==> "FCMH"

get a random sequence with the same NA count:

bioruby> counts = {'a'=>seq.count('a'),'c'=>seq.count('c'),'g'=>seq.count('g'),'t'=>seq.count('t')}
==> {"a"=>6, "c"=>2, "g"=>2, "t"=>2}
bioruby!> randomseq = Bio::Sequence::NA.randomize(counts) 
==!> "aaacatgaagtc"

bioruby!> print counts
a6c2g2t2  
bioruby!> p counts
{"a"=>6, "c"=>2, "g"=>2, "t"=>2}

The p, print and puts methods are standard Ruby ways of outputting to the screen. If you want to know more about standard Ruby commands you can use the 'ri' command on the command line (or the help command in Windows). For example

% ri puts
% ri p
% ri File.open

Nucleic acid sequence are members of the Bio::Sequence::NA class, and amino acid sequence are members of the Bio::Sequence::AA class. Shared methods are in the parent Bio::Sequence class.

As Bio::Sequence inherits Ruby's String class, you can use String class methods. For example, to get a subsequence, you can not only use subseq(from, to) but also String#[].

Please take note that the Ruby's string's are base 0 - i.e. the first letter has index 0, for example:

bioruby> s = 'abc'
==> "abc"
bioruby> s[0].chr
==> "a"
bioruby> s[0..1]
==> "ab"

So when using String methods, you should subtract 1 from positions conventionally used in biology. (subseq method will throw an exception if you specify positions smaller than or equal to 0 for either one of the "from" or "to".)

The window_search(window_size, step_size) method shows a typical Ruby way of writing concise and clear code using 'closures'. Each sliding window creates a subsequence which is supplied to the enclosed block through a variable named +s+.

Since the class of each subsequence is the same as original sequence (Bio::Sequence::NA or Bio::Sequence::AA or Bio::Sequence), you can use all methods on the subsequence. For example,

Finally, the window_search method returns the last leftover subsequence. This allows for example

If you don't want the overlapping window, set window size and stepping size to equal values.

Other examples

In most cases, sequences are read from files or retrieved from databases. For example:

require 'bio'

input_seq = ARGF.read       # reads all files in arguments

my_naseq = Bio::Sequence::NA.new(input_seq)
my_aaseq = my_naseq.translate

puts my_aaseq

Save the program above as na2aa.rb. Prepare a nucleic acid sequence described below and save it as my_naseq.txt:

gtggcgatctttccgaaagcgatgactggagcgaagaaccaaagcagtgacatttgtctg
atgccgcacgtaggcctgataagacgcggacagcgtcgcatcaggcatcttgtgcaaatg
tcggatgcggcgtga

na2aa.rb translates a nucleic acid sequence to a protein sequence. For example, translates my_naseq.txt:

% ruby na2aa.rb my_naseq.txt

or use a pipe!

% cat my_naseq.txt|ruby na2aa.rb

Outputs

VAIFPKAMTGAKNQSSDICLMPHVGLIRRGQRRIRHLVQMSDAA*

You can also write this, a bit fancifully, as a one-liner script.

% ruby -r bio -e 'p Bio::Sequence::NA.new($<.read).translate' my_naseq.txt

In the next section we will retrieve data from databases instead of using raw sequence files. One generic example of the above can be found in ./sample/na2aa.rb.

Parsing GenBank data (Bio::GenBank class)

We assume that you already have some GenBank data files. (If you don't, download some .seq files from ftp://ftp.ncbi.nih.gov/genbank/)

As an example we will fetch the ID, definition and sequence of each entry from the GenBank format and convert it to FASTA. This is also an example script in the BioRuby distribution.

A first attempt could be to use the Bio::GenBank class for reading in the data:

#!/usr/bin/env ruby

require 'bio'

# Read all lines from STDIN split by the GenBank delimiter
while entry = gets(Bio::GenBank::DELIMITER)
  gb = Bio::GenBank.new(entry)      # creates GenBank object

  print ">#{gb.accession} "         # Accession
  puts gb.definition                # Definition
  puts gb.naseq                     # Nucleic acid sequence 
                                    # (Bio::Sequence::NA object)
end

But that has the disadvantage the code is tied to GenBank input. A more generic method is to use Bio::FlatFile which allows you to use different input formats:

#!/usr/bin/env ruby

require 'bio'

ff = Bio::FlatFile.new(Bio::GenBank, ARGF)
ff.each_entry do |gb|
  definition = "#{gb.accession} #{gb.definition}"
  puts gb.naseq.to_fasta(definition, 60)
end

For example, in turn, reading FASTA format files:

#!/usr/bin/env ruby

require 'bio'

ff = Bio::FlatFile.new(Bio::FastaFormat, ARGF)
ff.each_entry do |f|
  puts "definition : " + f.definition
  puts "nalen      : " + f.nalen.to_s
  puts "naseq      : " + f.naseq
end

In the above two scripts, the first arguments of Bio::FlatFile.new are database classes of BioRuby. This is expanded on in a later section.

Again another option is to use the Bio::DB.open class:

#!/usr/bin/env ruby

require 'bio'

ff = Bio::GenBank.open("gbvrl1.seq")
ff.each_entry do |gb|
  definition = "#{gb.accession} #{gb.definition}"
  puts gb.naseq.to_fasta(definition, 60)
end

Next, we are going to parse the GenBank 'features', which is normally very complicated:

#!/usr/bin/env ruby

require 'bio'

ff = Bio::FlatFile.new(Bio::GenBank, ARGF)

# iterates over each GenBank entry
ff.each_entry do |gb|

  # shows accession and organism
  puts "# #{gb.accession} - #{gb.organism}"

  # iterates over each element in 'features'
  gb.features.each do |feature|
    position = feature.position
    hash = feature.assoc            # put into Hash

    # skips the entry if "/translation=" is not found
    next unless hash['translation']

    # collects gene name and so on and joins it into a string
    gene_info = [
      hash['gene'], hash['product'], hash['note'], hash['function']
    ].compact.join(', ')

    # shows nucleic acid sequence
    puts ">NA splicing('#{position}') : #{gene_info}"
    puts gb.naseq.splicing(position)

    # shows amino acid sequence translated from nucleic acid sequence
    puts ">AA translated by splicing('#{position}').translate"
    puts gb.naseq.splicing(position).translate

    # shows amino acid sequence in the database entry (/translation=)
    puts ">AA original translation"
    puts hash['translation']
  end
end

Bio::Sequence#splicing splices subsequences from nucleic acid sequences according to location information used in GenBank, EMBL and DDBJ.

When the specified translation table is different from the default (universal), or when the first codon is not "atg" or the protein contains selenocysteine, the two amino acid sequences will differ.

The Bio::Sequence#splicing method takes not only DDBJ/EMBL/GenBank feature style location text but also Bio::Locations object. For more information about location format and Bio::Locations class, see bio/location.rb.

You can also use this splicing method for amino acid sequences (Bio::Sequence::AA objects).

More databases

Databases in BioRuby are essentially accessed like that of GenBank with classes like Bio::GenBank, Bio::KEGG::GENES. A full list can be found in the ./lib/bio/db directory of the BioRuby source tree.

In many cases the Bio::DatabaseClass acts as a factory pattern and recognises the database type automatically - returning a parsed object. For example using Bio::FlatFile class as described above. The first argument of the Bio::FlatFile.new is database class name in BioRuby (such as Bio::GenBank, Bio::KEGG::GENES and so on).

ff = Bio::FlatFile.new(Bio::DatabaseClass, ARGF)

Isn't it wonderful that Bio::FlatFile automagically recognizes each database class?

#!/usr/bin/env ruby

require 'bio'

ff = Bio::FlatFile.auto(ARGF)
ff.each_entry do |entry|
  p entry.entry_id          # identifier of the entry
  p entry.definition        # definition of the entry
  p entry.seq               # sequence data of the entry
end

An example that can take any input, filter using a regular expression and output to a FASTA file can be found in sample/any2fasta.rb. With this technique it is possible to write a Unix type grep/sort pipe for sequence information. One example using scripts in the BIORUBY sample folder:

fastagrep.rb '/At|Dm/' database.seq | fastasort.rb

greps the database for Arabidopsis and Drosophila entries and sorts the output to FASTA.

Other methods to extract specific data from database objects can be different between databases, though some methods are common (see the guidelines for common methods in bio/db.rb).

Refer to the documents of each database to find the exact naming of the included methods.

In general, BioRuby uses the following conventions: when a method name is plural, the method returns some object as an Array. For example, some classes have a "references" method which returns multiple Bio::Reference objects as an Array. And some classes have a "reference" method which returns a single Bio::Reference object.

Alignments (Bio::Alignment)

The Bio::Alignment class in bio/alignment.rb is a container class like Ruby's Hash and Array classes and BioPerl's Bio::SimpleAlign. A very simple example is:

bioruby> seqs = [ 'atgca', 'aagca', 'acgca', 'acgcg' ]
bioruby> seqs = seqs.collect{ |x| Bio::Sequence::NA.new(x) }
# creates alignment object
bioruby> a = Bio::Alignment.new(seqs)
bioruby> a.consensus 
==> "a?gc?"
# shows IUPAC consensus
p a.consensus_iupac       # ==> "ahgcr"

# iterates over each seq
a.each { |x| p x }
  # ==>
  #    "atgca"
  #    "aagca"
  #    "acgca"
  #    "acgcg"
# iterates over each site
a.each_site { |x| p x }
  # ==>
  #    ["a", "a", "a", "a"]
  #    ["t", "a", "c", "c"]
  #    ["g", "g", "g", "g"]
  #    ["c", "c", "c", "c"]
  #    ["a", "a", "a", "g"]

# doing alignment by using CLUSTAL W.
# clustalw command must be installed.
factory = Bio::ClustalW.new
a2 = a.do_align(factory)

Read a ClustalW or Muscle 'ALN' alignment file:

bioruby> aln = Bio::ClustalW::Report.new(File.read('../test/data/clustalw/example1.aln'))
bioruby> aln.header
==> "CLUSTAL 2.0.9 multiple sequence alignment"

Fetch a sequence:

bioruby> seq = aln.get_sequence(1)
bioruby> seq.definition
==> "gi|115023|sp|P10425|"

Get a partial sequence:

bioruby> seq.to_s[60..120]
==> "LGYFNG-EAVPSNGLVLNTSKGLVLVDSSWDNKLTKELIEMVEKKFQKRVTDVIITHAHAD"

Show the full alignment residue match information for the sequences in the set:

bioruby> aln.match_line[60..120]
==> "     .     **. .   ..   ::*:       . * : : .        .: .* * *"

Return a Bio::Alignment object:

bioruby> aln.alignment.consensus[60..120]
==> "???????????SN?????????????D??????????L??????????????????H?H?D"

Restriction Enzymes (Bio::RE)

BioRuby has extensive support for restriction enzymes (REs). It contains a full library of commonly used REs (from REBASE) which can be used to cut single stranded RNA or double stranded DNA into fragments. To list all enzymes:

rebase = Bio::RestrictionEnzyme.rebase
rebase.each do |enzyme_name, info|
  p enzyme_name
end

and to cut a sequence with an enzyme follow up with:

res = seq.cut_with_enzyme('EcoRII', {:max_permutations => 0}, 
  {:view_ranges => true})
if res.kind_of? Symbol #error
   err = Err.find_by_code(res.to_s)
   unless err
     err = Err.new(:code => res.to_s)
   end
end
res.each do |frag|
   em = EnzymeMatch.new

   em.p_left = frag.p_left
   em.p_right = frag.p_right
   em.c_left = frag.c_left
   em.c_right = frag.c_right

   em.err = nil
   em.enzyme = ar_enz
   em.sequence = ar_seq
   p em
 end

Sequence homology search by using the FASTA program (Bio::Fasta)

Let's start with a query.pep file which contains a sequence in FASTA format. In this example we are going to execute a homology search from a remote internet site or on your local machine. Note that you can use the ssearch program instead of fasta when you use it in your local machine.

using FASTA in local machine

Install the fasta program on your machine (the command name looks like fasta34. FASTA can be downloaded from ftp://ftp.virginia.edu/pub/fasta/).

First, you must prepare your FASTA-formatted database sequence file target.pep and FASTA-formatted query.pep.

#!/usr/bin/env ruby

require 'bio'

# Creates FASTA factory object ("ssearch" instead of 
# "fasta34" can also work)
factory = Bio::Fasta.local('fasta34', ARGV.pop)
(EDITOR's NOTE: not consistent pop command)

ff = Bio::FlatFile.new(Bio::FastaFormat, ARGF)

# Iterates over each entry. the variable "entry" is a 
# Bio::FastaFormat object:
ff.each do |entry|
  # shows definition line (begins with '>') to the standard error output
  $stderr.puts "Searching ... " + entry.definition

  # executes homology search. Returns Bio::Fasta::Report object.
  report = factory.query(entry)

  # Iterates over each hit
  report.each do |hit|
    # If E-value is smaller than 0.0001
    if hit.evalue < 0.0001
      # shows identifier of query and hit, E-value, start and 
      # end positions of homologous region 
      print "#{hit.query_id} : evalue #{hit.evalue}\t#{hit.target_id} at "
      p hit.lap_at
    end
  end
end

We named above script f_search.rb. You can execute it as follows:

% ./f_search.rb query.pep target.pep > f_search.out

In above script, the variable "factory" is a factory object for executing FASTA many times easily. Instead of using Fasta#query method, Bio::Sequence#fasta method can be used.

seq = ">test seq\nYQVLEEIGRGSFGSVRKVIHIPTKKLLVRKDIKYGHMNSKE"
seq.fasta(factory)

When you want to add options to FASTA commands, you can set the third argument of the Bio::Fasta.local method. For example, the following sets ktup to 1 and gets a list of the top 10 hits:

factory = Bio::Fasta.local('fasta34', 'target.pep', '-b 10')
factory.ktup = 1

Bio::Fasta#query returns a Bio::Fasta::Report object. We can get almost all information described in FASTA report text with the Report object. For example, getting information for hits:

report.each do |hit|
  puts hit.evalue           # E-value
  puts hit.sw               # Smith-Waterman score (*)
  puts hit.identity         # % identity
  puts hit.overlap          # length of overlapping region
  puts hit.query_id         # identifier of query sequence
  puts hit.query_def        # definition(comment line) of query sequence
  puts hit.query_len        # length of query sequence
  puts hit.query_seq        # sequence of homologous region
  puts hit.target_id        # identifier of hit sequence
  puts hit.target_def       # definition(comment line) of hit sequence
  puts hit.target_len       # length of hit sequence
  puts hit.target_seq       # hit of homologous region of hit sequence
  puts hit.query_start      # start position of homologous 
                            # region in query sequence
  puts hit.query_end        # end position of homologous region 
                            # in query sequence
  puts hit.target_start     # start posiotion of homologous region 
                            # in hit(target) sequence
  puts hit.target_end       # end position of homologous region 
                            # in hit(target) sequence
  puts hit.lap_at           # array of above four numbers
end

Most of above methods are common to the Bio::Blast::Report described below. Please refer to the documentation of the Bio::Fasta::Report class for FASTA-specific details.

If you need the original output text of FASTA program you can use the "output" method of the factory object after the "query" method.

report = factory.query(entry)
puts factory.output

using FASTA from a remote internet site

For accessing a remote site the Bio::Fasta.remote method is used instead of Bio::Fasta.local. When using a remote method, the databases available may be limited, but, otherwise, you can do the same things as with a local method.

Available databases in GenomeNet:

Select the databases you require. Next, give the search program from the type of query sequence and database.

For example, run:

program = 'fasta'
database = 'genes'

factory = Bio::Fasta.remote(program, database)

and try out the same commands as with the local search shown earlier.

Homology search by using BLAST (Bio::Blast class)

The BLAST interface is very similar to that of FASTA and both local and remote execution are supported. Basically replace above examples Bio::Fasta with Bio::Blast!

For example the BLAST version of f_search.rb is:

# create BLAST factory object
factory = Bio::Blast.local('blastp', ARGV.pop)

For remote execution of BLAST in GenomeNet, Bio::Blast.remote is used. The parameter "program" is different from FASTA - as you can expect:

Bio::BLAST uses "-m 7" XML output of BLAST by default when either XMLParser or REXML (both of them are XML parser libraries for Ruby - of the two XMLParser is the fastest) is installed on your computer. In Ruby version 1.8.0 or later, REXML is bundled with Ruby's distribution.

When no XML parser library is present, Bio::BLAST uses "-m 8" tabular deliminated format. Available information is limited with the "-m 8" format so installing an XML parser is recommended.

Again, the methods in Bio::Fasta::Report and Bio::Blast::Report (and Bio::Fasta::Report::Hit and Bio::Blast::Report::Hit) are similar. There are some additional BLAST methods, for example, bit_score and midline.

report.each do |hit|
  puts hit.bit_score       
  puts hit.query_seq       
  puts hit.midline         
  puts hit.target_seq      

  puts hit.evalue          
  puts hit.identity        
  puts hit.overlap         
  puts hit.query_id        
  puts hit.query_def       
  puts hit.query_len       
  puts hit.target_id       
  puts hit.target_def      
  puts hit.target_len      
  puts hit.query_start     
  puts hit.query_end       
  puts hit.target_start    
  puts hit.target_end      
  puts hit.lap_at          
end

For simplicity and API compatibility, some information such as score is extracted from the first Hsp (High-scoring Segment Pair).

Check the documentation for Bio::Blast::Report to see what can be retrieved. For now suffice to say that Bio::Blast::Report has a hierarchical structure mirroring the general BLAST output stream:

See bio/appl/blast.rb and bio/appl/blast/*.rb for more information.

Parsing existing BLAST output files

When you already have BLAST output files and you want to parse them, you can directly create Bio::Blast::Report objects without the Bio::Blast factory object. For this purpose use Bio::Blast.reports, which supports the "-m 0" default and "-m 7" XML type output format.

Save the script as hits_under_0.001.rb and to process BLAST output files *.xml, you can run it with:

% ruby hits_under_0.001.rb *.xml

Sometimes BLAST XML output may be wrong and can not be parsed. Check whether blast is version 2.2.5 or later. See also blast --help.

Bio::Blast loads the full XML file into memory. If this causes a problem you can split the BLAST XML file into smaller chunks using XML-Twig. An example can be found in Biotools.

Add remote BLAST search sites

Note: this section is an advanced topic

Here a more advanced application for using BLAST sequence homology search services. BioRuby currently only supports GenomeNet. If you want to add other sites, you must write the following:

In addition, you must write a private class method in Bio::Blast named "exec_MYSITE" to get query sequence and to pass the result to Bio::Blast::Report.new(or Bio::Blast::Default::Report.new):

factory = Bio::Blast.remote(program, db, option, 'MYSITE')

When you write above routines, please send them to the BioRuby project, and they may be included in future releases.

Generate a reference list using PubMed (Bio::PubMed)

Nowadays using NCBI E-Utils is recommended. Use Bio::PubMed.esearch and Bio::PubMed.efetch.

#!/usr/bin/env ruby

require 'bio'

# NCBI announces that queries without email address will return error
# after June 2010. When you modify the script, please enter your email
# address instead of the staff's.
Bio::NCBI.default_email = 'staff@bioruby.org'

keywords = ARGV.join(' ')

options = {
  'maxdate' => '2003/05/31',
  'retmax' => 1000,
}

entries = Bio::PubMed.esearch(keywords, options)

Bio::PubMed.efetch(entries).each do |entry|
  medline = Bio::MEDLINE.new(entry)
  reference = medline.reference
  puts reference.bibtex
end

The script works same as pmsearch.rb. But, by using NCBI E-Utils, more options are available. For example published dates to search and maximum number of hits to show results can be specified.

See the help page of E-Utils for more details.

More about BibTeX

In this section, we explain the simple usage of TeX for the BibTeX format bibliography list collected by above scripts. For example, to save BibTeX format bibliography data to a file named genoinfo.bib.

% ./pmfetch.rb 10592173 >> genoinfo.bib
% ./pmsearch.rb genome bioinformatics >> genoinfo.bib

The BibTeX can be used with Tex or LaTeX to form bibliography information with your journal article. For more information on using BibTex see BibTex HowTo site. A quick example:

Save this to hoge.tex:

\documentclass{jarticle}
\begin{document}
\bibliographystyle{plain}
foo bar KEGG database~\cite{PMID:10592173} baz hoge fuga.
\bibliography{genoinfo}
\end{document}

Then,

% latex hoge
% bibtex hoge # processes genoinfo.bib
% latex hoge  # creates bibliography list
% latex hoge  # inserts correct bibliography reference

Now, you get hoge.dvi and hoge.ps - the latter of which can be viewed with any Postscript viewer.

Bio::Reference#bibitem

When you don't want to create a bib file, you can use Bio::Reference#bibitem method instead of Bio::Reference#bibtex. In the above pmfetch.rb and pmsearch.rb scripts, change

puts reference.bibtex

to

puts reference.bibitem

Output documents should be bundled in \begin{thebibliography} and \end{thebibliography}. Save the following to hoge.tex

\documentclass{jarticle}
\begin{document}
foo bar KEGG database~\cite{PMID:10592173} baz hoge fuga.

\begin{thebibliography}{00}

\bibitem{PMID:10592173}
Kanehisa, M., Goto, S.
KEGG: kyoto encyclopedia of genes and genomes.,
{\em Nucleic Acids Res}, 28(1):27--30, 2000.

\end{thebibliography}
\end{document}

and run

% latex hoge   # creates bibliography list
% latex hoge   # inserts corrent bibliography reference

OBDA

OBDA (Open Bio Database Access) is a standardized method of sequence database access developed by the Open Bioinformatics Foundation. It was created during the BioHackathon by BioPerl, BioJava, BioPython, BioRuby and other projects' members (2002).

This tutorial only gives a quick overview of OBDA. Check out the OBDA site for more extensive details.

BioRegistry

BioRegistry allows for locating retrieval methods and database locations through configuration files. The priorities are

Note that the last locaation refers to www.open-bio.org and is only used when all local configulation files are not available.

In the current BioRuby implementation all local configulation files are read. For databases with the same name settings encountered first are used. This means that if you don't like some settings of a database in the system's global configuration file (/etc/bioinformatics/seqdatabase.ini), you can easily override them by writing settings to ~/.bioinformatics/seqdatabase.ini.

The syntax of the configuration file is called a stanza format. For example

[DatabaseName]
protocol=ProtocolName
location=ServerName

You can write a description like the above entry for every database.

The database name is a local label for yourself, so you can name it freely and it can differ from the name of the actual databases. In the actual specification of BioRegistry where there are two or more settings for a database of the same name, it is proposed that connection to the database is tried sequentially with the order written in configuration files. However, this has not (yet) been implemented in BioRuby.

In addition, for some protocols, you must set additional options other than locations (e.g. user name for MySQL). In the BioRegistory specification, current available protocols are:

In BioRuby, you can use index-flat, index-berkleydb, biofetch and biosql. Note that the BioRegistry specification sometimes gets updated and BioRuby does not always follow quickly.

Here is an example. It creates a Bio::Registry object and reads the configuration files:

reg = Bio::Registry.new

# connects to the database "genbank"
serv = reg.get_database('genbank')

# gets entry of the ID
entry = serv.get_by_id('AA2CG')

The variable "serv" is a server object corresponding to the settings written in the configuration files. The class of the object is one of Bio::SQL, Bio::Fetch, and so on. Note that Bio::Registry#get_database("name") returns nil if no database is found.

After that, you can use the get_by_id method and some specific methods. Please refer to the sections below for more information.

BioFlat

BioFlat is a mechanism to create index files of flat files and to retrieve these entries fast. There are two index types. index-flat is a simple index performing binary search without using any external libraries of Ruby. index-berkeleydb uses Berkeley DB for indexing - but requires installing bdb on your computer, as well as the BDB Ruby package. To create the index itself, you can use br_bioflat.rb command bundled with BioRuby.

% br_bioflat.rb --makeindex database_name [--format data_format] filename...

The format can be omitted because BioRuby has autodetection. If that doesn't work, you can try specifying the data format as the name of a BioRuby database class.

Search and retrieve data from database:

% br_bioflat.rb database_name identifier

For example, to create an index of GenBank files gbbct*.seq and get the entry from the database:

% br_bioflat.rb --makeindex my_bctdb --format GenBank gbbct*.seq
% br_bioflat.rb my_bctdb A16STM262

If you have Berkeley DB on your system and installed the bdb extension module of Ruby (see the BDB project page ), you can create and search indexes with Berkeley DB - a very fast alternative that uses little computer memory. When creating the index, use the "--makeindex-bdb" option instead of "--makeindex".

% br_bioflat.rb --makeindex-bdb database_name [--format data_format] filename...

BioFetch

Note: this section is an advanced topic

BioFetch is a database retrieval mechanism via CGI. CGI Parameters, options and error codes are standardized. Client access via http is possible giving the database name, identifiers and format to retrieve entries.

The BioRuby project has a BioFetch server at bioruby.org. It uses GenomeNet's DBGET system as a backend. The source code of the server is in sample/ directory. Currently, there are only two BioFetch servers in the world: bioruby.org and EBI.

Here are some methods to retrieve entries from our BioFetch server.

  1. Using a web browser

    http://bioruby.org/cgi-bin/biofetch.rb
  2. Using the br_biofetch.rb command

    % br_biofetch.rb db_name entry_id
  3. Directly using Bio::Fetch in a script

    serv = Bio::Fetch.new(server_url)
    entry = serv.fetch(db_name, entry_id)
  4. Indirectly using Bio::Fetch via BioRegistry in script

    reg = Bio::Registry.new
    serv = reg.get_database('genbank')
    entry = serv.get_by_id('AA2CG')

If you want to use (4), you have to include some settings in seqdatabase.ini. For example:

[genbank]
protocol=biofetch
location=http://bioruby.org/cgi-bin/biofetch.rb
biodbname=genbank

The combination of BioFetch, Bio::KEGG::GENES and Bio::AAindex1

Bioinformatics is often about gluing things together. Here is an example that gets the bacteriorhodopsin gene (VNG1467G) of the archaea Halobacterium from KEGG GENES database and gets alpha-helix index data (BURA740101) from the AAindex (Amino acid indices and similarity matrices) database, and shows the helix score for each 15-aa length overlapping window.

#!/usr/bin/env ruby

require 'bio'

entry = Bio::Fetch.query('hal', 'VNG1467G')
aaseq = Bio::KEGG::GENES.new(entry).aaseq

entry = Bio::Fetch.query('aax1', 'BURA740101')
helix = Bio::AAindex1.new(entry).index

position = 1
win_size = 15

aaseq.window_search(win_size) do |subseq|
  score = subseq.total(helix)
  puts [ position, score ].join("\t")
  position += 1
end

The special method Bio::Fetch.query uses the preset BioFetch server at bioruby.org. (The server internally gets data from GenomeNet. Because the KEGG/GENES database and AAindex database are not available from other BioFetch servers, we used the bioruby.org server with Bio::Fetch.query method.)

BioSQL

BioSQL is a well known schema to store and retrive biological sequences using a RDBMS like PostgreSQL or MySQL: note that SQLite is not supported. First of all, you must install a database engine or have access to a remote one. Then create the schema and populate with the taxonomy. You can follow the Official Guide to accomplish these steps. Next step is to install these gems:

You can find ActiveRecord's models in /bioruby/lib/bio/io/biosql

When you have your database up and running, you can connect to it like this:

#!/usr/bin/env ruby

require 'bio'

connection = Bio::SQL.establish_connection({'development'=>{'hostname'=>"YourHostname",
'database'=>"CoolBioSeqDB",
'adapter'=>"jdbcmysql",
'username'=>"YourUser",
'password'=>"YouPassword"
      }
  },
'development')

#The first parameter is the hash contaning the description of the configuration; similar to database.yml in Rails applications, you can declare different environment. 
#The second parameter is the environment to use: 'development', 'test', or 'production'.

#To store a sequence into the database you simply need a biosequence object.
biosql_database = Bio::SQL::Biodatabase.find(:first)
ff = Bio::GenBank.open("gbvrl1.seq")

ff.each_entry do |gb|
  Bio::SQL::Sequence.new(:biosequence=>gb.to_biosequence, :biodatabase=>biosql_database
end

#You can list all the entries into every database 
Bio::SQL.list_entries

#list databases:
Bio::SQL.list_databases

#retriving a generic accession
bioseq = Bio::SQL.fetch_accession("YouAccession")

#If you use biosequence objects, you will find all its method mapped to BioSQL sequences. 
#But you can also access to the models directly:

#get the raw sequence associated with your accession
bioseq.entry.biosequence 

#get the length of your sequence; this is the explicit form of bioseq.length
bioseq.entry.biosequence.length

#convert the sequence into GenBank format
bioseq.to_biosequence.output(:genbank)

BioSQL's schema is not very intuitive for beginners, so spend some time on understanding it. In the end if you know a little bit of Ruby on Rails, everything will go smoothly. You can find information on Annotation here. ToDo: add exemaples from George. I remember he did some cool post on BioSQL and Rails.

PhyloXML

PhyloXML is an XML language for saving, analyzing and exchanging data of annotated phylogenetic trees. PhyloXML's parser in BioRuby is implemented in Bio::PhyloXML::Parser, and its writer in Bio::PhyloXML::Writer. More information can be found at www.phyloxml.org.

Bio::PhyloXML have been split out from BioRuby core and have been released as bio-phyloxml gem. To use Bio::PhyloXML, install the bio-phyloxml gem.

% gem install bio-phyloxml

The tutorial of Bio::PhyloXML is bundled in bio-phyloxml. <URL:https://github.com/bioruby/bioruby-phyloxml/blob/master/doc/Tutorial.rd>

The BioRuby example programs

Some sample programs are stored in ./samples/ directory. For example, the n2aa.rb program (transforms a nucleic acid sequence into an amino acid sequence) can be run using:

./sample/na2aa.rb test/data/fasta/example1.txt 

Unit testing and doctests

BioRuby comes with an extensive testing framework with over 1300 tests and 2700 assertions. To run the unit tests:

cd test
ruby runner.rb

We have also started with doctest for Ruby. We are porting the examples in this tutorial to doctest - more info upcoming.

Further reading

See the BioRuby in anger Wiki. A lot of BioRuby's documentation exists in the source code and unit tests. To really dive in you will need the latest source code tree. The embedded rdoc documentation for the BioRuby source code can be viewed online at <URL:http://bioruby.org/rdoc/>.

BioRuby Shell

The BioRuby shell implementation is located in ./lib/bio/shell. It is very interesting as it uses IRB (the Ruby intepreter) which is a powerful environment described in Programming Ruby's IRB chapter. IRB commands can be typed directly into the shell, e.g.

bioruby!> IRB.conf[:PROMPT_MODE]
==!> :PROMPT_C

Additionally, you also may want to install the optional Ruby readline support - with Debian libreadline-ruby. To edit a previous line you may have to press line down (down arrow) first.

Helpful tools

Apart from rdoc you may also want to use rtags - which allows jumping around source code by clicking on class and method names.

cd bioruby/lib
rtags -R --vi

For a tutorial see here

APPENDIX

Biogem: Additional BioRuby plugins

Biogem is one of the exciting developments for Ruby in bioinformatics! Biogems add new functionality next to the BioRuby core project (BioRuby is a biogem itself). A biogem is simply installed with

gem install bio                 # The core BioRuby gem
gem install bio-core            # BioRuby + stable pure Ruby biogems
gem install bio-core-ext        # bio-core + stable Ruby extensions

Information on these biogems, and the many others available, see Biogems.info or gems.bioruby.org.

Ruby Ensembl API

The Ruby Ensembl API is a Ruby API to the Ensembl database. It is NOT currently included in the BioRuby archives. To install it, see the Ruby-Ensembl Github for more information.

Gene Ontology (GO) through the Ruby Ensembl API

Gene Ontologies can be fetched through the Ruby Ensembl API package:

require 'ensembl'
Ensembl::Core::DBConnection.connect('drosophila_melanogaster')
infile = IO.readlines(ARGV.shift) # reading your comma-separated accession mapping file (one line per mapping)
infile.each do |line|
  accs = line.split(",")          # Split the comma-sep.entries into an array
  drosphila_acc = accs.shift      # the first entry is the Drosophila acc
  mosq_acc = accs.shift           # the second entry is your Mosq. acc
  gene = Ensembl::Core::Gene.find_by_stable_id(drosophila_acc)
  print "#{mosq_acc}"
  gene.go_terms.each do |go|
     print ",#{go}"
  end
end

Prints each mosq. accession/uniq identifier and the GO terms from the Drosphila homologues.

Using BioPerl or BioPython from Ruby

A possible route is to opt for JRuby and Jython on the JAVA virtual machine (JVM).

At the moment there is no easy way of accessing BioPerl or BioPython directly from Ruby. A possibility is to create a Perl or Python server that gets accessed through XML/RPC or SOAP.

Installing required external libraries

At this point for using BioRuby no additional libraries are needed.

This may change, so keep an eye on the Bioruby website. Also when a package is missing BioRuby should show an informative message.

At this point installing third party Ruby packages can be a bit painful, as the gem standard for packages evolved late and some still force you to copy things by hand. Therefore read the README's carefully that come with each package.

Trouble shooting

Ruby is failing to find the BioRuby libraries - add it to the RUBYLIB path, or pass it to the interpeter. For example:

ruby -I$BIORUBYPATH/lib yourprogram.rb

Modifying this page

IMPORTANT NOTICE: This page is maintained in the BioRuby source code repository. Please edit the file there otherwise changes may get lost. See BioRuby Developer Information for repository and mailing list access.