1. Comparative Data Analysis Ontology (CDAO)
Francisco Prosdocimi, Brandon Chisham, Julie Thompson, Enrico Pontelli, Arlin Stoltzfus

2. Objectives
Develop a framework to formalize knowledge in the evolutionary biology domain
Formalize an ontology for comparative data analysis
Comparative Data Analysis Ontology (CDAO)
Implement and Evaluate the ontology

3. Motivation Interoperation Reasoning Miscellaneous
Ontologies formalize knowledge
Overcome ambiguities in data formats (e.g., the multiple interpretations of NEXUS)
Facilitate provably correct format conversions
Reasoning
Beyond relational queries
Automated generation of format converters
Advanced reasoning required for workflow constructions and validation
Miscellaneous
Guide development of new data formats
Lingua franca for knowledge exchange …

4. Choice of Representation
Development Process
Ontology Construction
Conceptualization
Concept glossary
Organize knowledge
Implementation
Build model
Integrate ontologies
Evaluation
Consistency tests
Completeness tests
Use cases
Specification
Identify purpose and scope
Define use cases
Choice of Representation
Language
Development tools

5. Structure of CDAO Current Focus: Taxonomic units
Tree-like networks of relationships
Models of evolutionary changes

6. Structure of CDAO Core Components Imported Components
Representation of Networks and Trees (e.g., NEXUS TREE Block)
Representation of Character Data (e.g., NEXUS CHARACTERS Block)
Imported Components
Amino Acid Ontology
U. Manchester, 2006
Nucleotide Ontology

7. CDAO: Core Components Network/Tree representation
Rooted and Unrooted Trees
Nodes
Edges
Sets of Nodes
Child Node
topology
node
rooted tree
part_of
directed edge
has ancestor
is_a
has descendant
network
node
is_a
Parent Node
is_a
Unrooted tree
part_of
edge
node
Represents TU
part_of
has_annotation
has_annotation
has element
mrca_of
Annotation: Tree Procedure, Model…
Annotation: Transformation, Length…
set of nodes
is_a
lineage

8. has_descendant min 2 Nodes
CDAO: Core Components
Representation of a Directed Tree a) A B C D E
has_descendant min 2 Nodes
Lineage
Subtree
MRCA_Node
has_child_Node
Directed edge or branch
has_parent_Node
Edge Transformation
has_root_node
Character
Ancestor state,
Derived state…
Edge
Node
Node (Ancestral)
Rooted_tree
Edge Transformation

9. CDAO: Core Components Annotations Edge Annotations Model Description
Length
Transformation
Model Description
Gap Cost
Substitution Model
TU Annotation
Taxonomic Link
Tree Annotation
Tree Procedure
Edge Annotation
transform_character
has_left_state
has_left_node
character state
transformation
character state
has_right_node
has_right_state

10. CDAO: Core Components Character State Data Matrix Character
Taxonomic Units
Datum
State
Character State Data Matrix
has annotation
Annotation: Alignment procedures…
character state data matrix
part_of
part_of
Annotation: TAXID, DB-XREF…
has annotation
belongs_to
taxonomic unit
has datum
character
has datum
character state datum
belongs_to
has
represented by node
has
coordinate
character state
is_a
is_a
belongs_to
is_a
is_a
compound
amino acid
discrete
coordinate system
is_transformation_of
is_a
nucleotide
continuous

11. Implementation Details
Formalization
OWL 1.1
Tools
Protégé 4 [edit]
Swoop 2.3 [validation]
C++ and Perl+Prolog translators
Swoop 2.3 [reasoning]
Pellet [reasoning]
Fact++ [reasoning]

12. Preliminary Evaluation
We are reaching the stage where concrete evaluation is possible
NEXUS converters
We stumbled on several blocks
A good formalization of CDAO requires sophisticated features (OWL 1.1)
The majority of reasoning engines has not reached OWL 1.1 yet (even if they claim so…)

13. Some Examples Simple NEXUS file #NEXUS BEGIN TAXA; DIMENSIONS ntax=10;
TAXLABELS Arabidopsis_thaliana_AAD Arabidopsis_thaliana_CAB Oryza_sativa_BAB Dictyostelium_discoideum_AAO Caenorhabditis_elegans_CAA Drosophila_melanogaster_AAF Drosophila_melanogaster_AAF Mus_musculus_BAB Saccharomyces_cerevisiae_AAB Schizosaccharomyces_pombe_CAB ;
END;
BEGIN CHARACTERS;
TITLE dna;
LINK taxa=PF00137_47;
DIMENSIONS nchar=10;
FORMAT datatype=dna gap=- missing=?;
MATRIX
Arabidopsis_thaliana_CAB gtgtggttgc
Schizosaccharomyces_pombe_CAB tgtatatgct
Drosophila_melanogaster_AAF tgtacttcgt
Arabidopsis_thaliana_AAD gt---gtggc
Oryza_sativa_BAB ct
Saccharomyces_cerevisiae_AAB tgtacaagct
Mus_musculus_BAB tctgctacac
Dictyostelium_discoideum_AAO cacttactcc
Caenorhabditis_elegans_CAA tgttttacat
Drosophila_melanogaster_AAF ac------g- ;
BEGIN TREES;
TREE con_50_majrule = (((Arabidopsis_thaliana_AAD : ,Arabidopsis_thaliana_CAB : )inode15: ,Oryza_sativa_BAB : )inode14: ,(Dictyostelium_discoideum_AAO : ,(((Caenorhabditis_elegans_CAA : ,(Drosophila_melanogaster_AAF : ,Drosophila_melanogaster_AAF : )inode20: )inode19: ,Mus_musculus_BAB : )inode18: ,(Saccharomyces_cerevisiae_AAB : ,Schizosaccharomyces_pombe_CAB : )inode21: )inode17: )inode16: )root;

14. Some Examples Node: Directed_Edge:
<cdao:Node rdf:ID="node_inode15">
<cdao:part_of rdf:resource="#Tree"/>
<cdao:belongs_to_Edge rdf:resource="#edge_inode15_inode14" />
<cdao:belongs_to_Edge rdf:resource="#edge_Arabidopsis_thaliana_CAB79970_1_inode15" />
<cdao:belongs_to_Edge rdf:resource="#edge_Arabidopsis_thaliana_AAD31363_1_inode15" />
<cdao:belongs_to_Edge_as_Child rdf:resource="#edge_inode15_inode14" />
<cdao:belongs_to_Edge_as_Parent rdf:resource="#edge_Arabidopsis_thaliana_CAB79970_1_inode15" />
<cdao:belongs_to_Edge_as_Parent rdf:resource="#edge_Arabidopsis_thaliana_AAD31363_1_inode15" />
<cdao:nca_node_of rdf:resource="#set_nca_44"/>
</cdao:Node>
Directed_Edge:
<cdao:Directed_Edge rdf:ID="edge_Arabidopsis_thaliana_CAB79970_1_inode15">
<cdao:has_Parent_Node rdf:resource="#node_inode15"/>
<cdao:has_Child_Node rdf:resource="#node_Arabidopsis_thaliana_CAB79970_1"/>
<cdao:has_Annotation rdf:resource="#edge_Arabidopsis_thaliana_CAB79970_1_inode15_length"/>
</cdao:Directed_Edge>
<cdao:Edge_Length rdf:ID="edge_Arabidopsis_thaliana_CAB79970_1_inode15_length">
<cdao:has_Value rdf:datatype="&xsd;float"> </cdao:has_Value>
</cdao:Edge_Length>

15. Some Examples TU Character Datum State
<cdao:TU rdf:ID="Caenorhabditis_elegans_CAA92686_1">
<cdao:belongs_to_Character_State_Data_Matrix rdf:resource="#Matrix"/>
<cdao:represented_by_Node rdf:resource="#node_Caenorhabditis_elegans_CAA92686_1"/>
<cdao:has_Nucleotide_Datum rdf:resource="#datum_Caenorhabditis_elegans_CAA92686_1_char_0"/>
<cdao:has_Nucleotide_Datum rdf:resource="#datum_Caenorhabditis_elegans_CAA92686_1_char_1"/>
<cdao:has_Nucleotide_Datum rdf:resource="#datum_Caenorhabditis_elegans_CAA92686_1_char_2"/> …
</cdao:TU>
Character
<cdao:Nucleotide_Character rdf:ID="char_2">
<cdao:has_Nucleotide_Datum rdf:resource="#datum_Oryza_sativa_BAB21282_1_char_2"/>
<cdao:has_Nucleotide_Datum rdf:resource="#datum_Arabidopsis_thaliana_CAB79970_1_char_2"/>
<cdao:has_Nucleotide_Datum rdf:resource="#datum_Mus_musculus_BAB61955_1_char_2"/>
</cdao:Nucleotide_Character>
Datum
<cdao:Nucleotide_State_Datum rdf:ID="datum_Caenorhabditis_elegans_CAA92686_1_char_6">
<cdao:belongs_to_Character rdf:resource="#char_6"/>
<cdao:belongs_to_TU rdf:resource="#Caenorhabditis_elegans_CAA92686_1"/>
<cdao:has_Nucleotide_State rdf:resource="#value_a"/>
</cdao:Nucleotide_State_Datum>
State
<cdao:Nucleotide rdf:ID="value_a">
<owl:sameAs rdf:resource="#dA"/>
</cdao:Nucleotide>

16. Simple Reasoning Tasks
Determine what TUs contain a gap in their tables: [Fact++]
(has_Datum some (has_State value gap)) and TU
Determine the ancestors of a TU in the tree:
has_Descendant value node_Drosophila_melanogaster_AAF55115_1

17. Simple Reasoning Tasks
Extract the row of a specific TU:
SELECT ?z,?y WHERE (base:Arabidopsis_thaliana_AAD31363_1>, cdao:has_Datum, ?x) (?x, cdao:has_State, ?y) (?x, cdao:belongs_to_Character, ?z) USING base FOR <file:/C:/Users/epontell/Documents/Research/Proposals/NEXUS/Research/Perl/inst_matrix.owl#>, cdao FOR < z y
inst_matrix:char_9
CURRENT:dC
inst_matrix:char_8
CURRENT:dG
inst_matrix:char_7
inst_matrix:char_5
inst_matrix:char_0
inst_matrix:char_6
CURRENT:dT
inst_matrix:char_1
inst_matrix:char_3
CURRENT:gap
inst_matrix:char_2
inst_matrix:char_4

18. Future Work To facilitate evaluation Java-level reasoning
Create an OWL 1.0 edition of the ontology (and corresponding NEXUS translator)
Java-level reasoning
Aggregation
Etc.
Large scale NEXUS validation
NeXML Interface
OBO distribution