Difference between revisions of "NeXML and RDF API for BioRuby"

From Phyloinformatics
Jump to: navigation, search
m (Serializing)
Line 47: Line 47:
   # => "<otus label=\"A taxa block\" id=\"taxa1\">\n  <otu label=\"A taxon\" id=\"o1\"/>\n</otus>"
   # => "<otus label=\"A taxa block\" id=\"taxa1\">\n  <otu label=\"A taxon\" id=\"o1\"/>\n</otus>"
Unit tests for serializer are filled with such use cases.

Revision as of 16:57, 30 June 2010


The following document discusses the implementation of an NeXML parser and serializer and an RDF API for BioRuby. Note that this document is not final yet.


Currently all the parsing is done at the start( i.e. no streaming ). This is likely to change later. Parse an NeXML file:

 doc = Bio::NeXML::Parser.new( "trees.xml" )
 nexml = doc.parse
 nexml.class #Bio::NeXML::Nexml


Bio::NeXML::Writer class provides a wrapper over libxml-ruby to create any NeXML document. This class defines a set of serialize_* instance methods which can be called on the appropriate object to get its NeXML representation. The method returns a XML::Node object. To get the raw NeXML representation to_s method should be called on the return value.

NeXML defines three top level containers: otus, trees, characters which bear parent-child relation with other NeXML elements. In effect, a valid NeXML document has only three type of immediate children. Naturally, a typical working paradigm would be to create Bio::NeXML::Otus, Bio::NeXML::Trees, and Bio::NeXML::Characters objects and write them to the NeXML file.

 # Parse a test file. This will give us Bio::NeXML::Otus,
 # Bio::NeXML::Trees, and Bio::NeXML::Characters object.
 doc1 = Bio::NeXML::Parser.new 'test.xml'
 nexml = doc1.parse
 # Create a Writer object,
 writer = Bio::NeXML::Writer.new
 # add otus, trees and characters to it,
 writer << nexml.otus
 writer << nexml.trees
 writer << nexml.characters
 # save it.
 writer.save 'sample.xml'

Bio::NeXML::Writer#<< internally calls some serialize_* method at the lowest level. If need be, these serialize_* methods can be called to obtain raw NeXML representation of any NeXML element.

 # Create an otus object with a child otu element
 taxa1 = Bio::NeXML::Otus.new 'taxa1', 'A taxa block'
 o1 = Bio::NeXML::Otu.new 'o1', 'A taxon'
 taxa1 << o1
 # Obtain the raw NeXML representation of the otus object created
 writer = Bio::NeXML::Writer.new
 writer.serialize_otus( taxa1 ).to_s
 # => "<otus label=\"A taxa block\" id=\"taxa1\">\n  <otu label=\"A taxon\" id=\"o1\"/>\n</otus>"

Unit tests for serializer are filled with such use cases.


 #get a hash of otus objects indexed with 'id'
 #get an array of otus objects
 #get an otus by id
 taxa1 = nexml.get_otus_by_id "taxa1"
 #iterate over each otus object
 nexml.each_otus do |taxa|
   puts taxa.id
   puts taxa.label
 nexml.trees_set #return a hash of trees object indexed with 'id'
 nexml.trees #return an array of trees objects.
 #iterate over each trees object
 nexml.each_trees do |trees|
   puts trees.id
   puts trees.label
 #find a trees by id
 trees1 = nexml.get_trees_by_id 'trees1'
 # characters
 nexml.characters_set #return a hash of characters object indexed with 'id'
 nexml.characters #return an array of characters object
 #iterate over each characters object
 nexml.each_characters do |ch|
   puts ch.id
   puts ch.label
 #find a characters object by id
 characters = nexml.get_characters_by_id 'chars1'


Taxa blocks and taxons are stored internally as a Ruby hash for faster 'id' based lookup. Consider this[1] NeXML snippet

 #get the id of otus
 taxa1.id # "taxa1"
 #get the label of otus
 taxa1.label # "Primary taxa block"
 #get a hash of child otu objects indexed with id
 #get an array of child otu objects
 #get an otu object by id
 #get_otu_by_id is an alias of []
 t1 = taxa1[ 't1' ]
 #add an otu object to otus
 t1.add_otu( otu_object )
 #to add more than one otu object at a time use << or otus= method
 t1 << [otu_object1, otu_object2]
 t1.otus = otu_object1, otu_object2
 #or iterate over each otu object
 #each_otu is an alias for each
 taxa1.each do |taxon|
   puts taxon.id
   puts taxon.label
 #check if an otu with given id belongs to an otus or not
 #include? and has? are alias for has_otu?
 taxa1.has_otu? 't2' # => true
 taxa1.has? 't8' # => false
 #an otus object in enumerable
 taxa1.map &:id # => array of otu ids
 taxa1.select {|t| t.class == "Lemurs" } #maybe in future


 #get an otu's id
 t1.id # => "t1"
 #get an otu's label
 t1.label # => "Homo sapiens"


Trees and tree and network are stored internally as a Ruby hash for faster 'id' based lookup.

 trees1.class #Bio::NeXML::Trees
 #get the taxa block to which the trees is linked to
 trees1.otus #returns an otus object


 trees1.tree_set #return a hash or tree objects indexed with 'id'
 tress1.trees #return an arrayof trees object
 #iterate over each tree object
 trees1.each_tree do |t|
   puts t.id
   puts t.label
 #get a tree object with its 'tree1'
 tree1 = trees1[ 'tree1' ]
 #or, with a conventional method call
 tree1 = trees1.get_tree_by_id 'tree1'
 #or, from a nexml object
 tree1 = nexml.get_tree_by_id 'tree1'
 tree1.class #Bio::NeXML::IntTree or Bio::NeXML::FloatTree
 #check if a tree belongs to a trees or not
 #pass it a tree id
 tree1.has_tree? 'tree1' #return true or false
 #get the number of treess


 trees1.network_set #return a hash or network objects indexed with 'id'
 tress1.networks #return an arrayof network objects
 #iterate over each network object
 trees1.each_network do |n|
   puts n.id
   puts n.label
 #get a network object with its id
 network1 = trees1[ 'network1' ]
 #or, with a conventional method call
 network1 = trees1.get_network_by_id 'network1'
 #or, from a nexml object
 network1 = nexml.get_tree_by_id 'network1'
 network1.class #Bio::NeXML::IntTree or Bio::NeXML::FloatTree
 #check if a network belongs to a trees or not
 #pass it a network id
 trees1.has_network? 'network1' #return true or false
 #get the number of networks

Tree and Network:

 #iterate over both trees and networks
 trees1.each do |g|
   puts g.class
 #find if a tree or a network belongs to a trees or not
 #include? is an alias for has?
 trees1.has? 'tree1' #return true or false
 #total number of trees and networks

All the available methods from Bio::Tree class can be called on a tree object.

 node1 = tree.get_node_by_name "n3" #note name is same as id
 tree1.parents node1

A trees object is an enumerable:

 trees1.map &:id


 puts characters.class
 #get the taxa block to which the characters is linked to
 characters.otus #returns an otus object
 #get the child format element
 format = characters.format
 puts format.class
 #get the child matrix element
 matrix = characters.matrix
 puts matrix.class


 format.states_set #return a hash of states objects indexed with 'id'
 format.states #return an array of states object
 #iterate over each states object
 format.each_states do |states|
   puts states.id
   puts states.label
 #get a states object by id
 states = format.get_states_by_id 'states1'
 #check if the states object with 'id' belongs to format or not
 format.has_states? 'states1'
 format.char_set #return a hash of char objects indexed with 'id'
 format.chars #return an array of char objects
 #iterate over each char object
 format.each_char do |char|
   puts char.id
   puts char.label
 #get a char object by id
 char = format.get_char_by_id 'char1'
 #check if the char object with 'id' belongs to format or not
 format.has_char? 'char1'
 #get a states or a char object by id
 state = format[ 'states1' ]
 char = format[ 'char1' ]
 #check if a states or a char object with 'id' belongs to format or not
 format.has? 'states1'
 format.has? 'char1'
 #all objects, including char and states can be iterated over with each
 format.each do |obj|
   puts obj.class
 #format is enumerable
 format.map &:id


 states.state_set #return a hash of state objects indexed with 'id'
 states.states #return an array of state objects
 #iterate over each state object
 states.each_state do |state|
   puts state.id
 #or, use its alias each
 #get a state object by id
 state = states.get_state_by_id 'state1'
 #or, use hash notation
 state = states[ 'state1' ]
 #check if a state belongs to states or not
 states.has_state? 'state1'
 #or, use its alias has? and include?


 #get the symbol associated with the state
 #find if the state is ambiguous
 #find the kind of ambiguity
 #find if it is an uncertain state set
 #find if it is a polymorphic state set
 #get the members of a state set as an array
 #or iterate over each member
 state.each do |member|
   puts member.class #same as self
   puts member.id
 #a state is Enumerable over its members
 state.select{ |member| member.id == "rna5" }


 #get the id
 #get the label
 #get the states object the char is linked to
 #get the codon position for DnaChar and RnaChar objects