Data processing is at the heart of science. Hence, the problem of data typing is an important problem that has been addressed from different aspects and in different forms. For example, the Research Data Alliance1 (RDA), whose major goal is to speed up the international data-driven innovation and discovery by facilitating research data sharing and exchange, has identified that the problem of data typing is an important problem that deserves attention. For this purpose, the RDA formed a Data Type Registry (DTR) working group with the goal to: compile a set of use cases for datatype use and management, formulate a data model and expression for datatypes, design a functional specification for type registries, and propose a federation strategy among multiple type registries. In data mining research it is impossible to efficiently connect parts of workflows (semi-) automatically, such as data pre-processing and data mining, perform analysis of the research results and communicate the research outputs, without machine-processable representation of datatypes and their properties. Hence, there is a need for a standardized semantically-defined and machine amenable representation of scientific datatypes to support crossdomain applications. Unfortunately, the existing representations of datatypes do not fully address such a need. To address this gap, we built an generic ontology for the representation of scientific knowledge about datatypes, named OntoDT (Panov et al., 2015). 1 http://rd-alliance.org/ 2 http://tinyurl.com/qdua9f7 3 http://tinyurl.com/nmjnlw2
OntoDT defines the meaning of the key entities and represents the knowledge about datatypes in a machine friendly way. The OntoDT ontology is based on the latest revised version of the ISO/IEC 114042 standard for datatypes. The design of the OntoDT ontology follows best practices in ontology engineering, such as the OBO Foundry principles. We used the Information Artifact Ontology3 (IAO) to define the upper level classes and re-used existing ontological resources, such as Open Biomedical Ontologies.
The OntoDT ontology defines the basic entities (see Fig. 1), such as datatype, properties of datatypes, value space, and characterizing operations. We also define a taxonomy of datatypes. The top-level ontology classes include primitive datatypes, generated datatypes, and user defined datatypes. Primitive datatypes are defined by explicit specification and are independent of other datatypes. Generated datatypes are syntactically and semantically dependent on other datatypes, and are specified implicitly with datatype generators. User defined datatypes are defined by a datatype declaration and allow defining additional identifiers and refinements to both primitive and generated datatypes. At the lower levels, the datatypes are distinguished with respect to their datatype properties.
OntoDT was used within an Ontology of core data mining
entities for constructing taxonomies of datasets, data mining tasks,
generalizations and data mining algorithms
OntoDT is a generic ontology and it allows easy extensions to represent domain specific datatypes. This can be done by directly extending the OntoDT datatype taxonomy and defining the semantic meaning of the domain datatypes by linking them to the corresponding entities in other domain ontologies. For example, we can define an amino-acid sequence datatype as a subclass of the character sequence datatype class (which is a sequence datatype having characters as its base type). Its semantic meaning can be defined via the IS-ABOUT relation to the amino acid sequence entity Panov et al provided by the National Cancer Institute Thesaurus4. In this way, OntoDT can be used for representation of bioinformatics datatypes.
Currently, BioXSD is used to define the basic bio-informatics
types of data (
For example, the BioXSD datatype sequence represents a string of 1-letter coded nucleotides or amino-acids. A sequence record is a datatype containing a sequence, and optionally some metadata about the sequence (for the purpose of identification). The semantic meanings of the terms sequence and nucleotide are curtail for the capturing of the semantic meaning of the data of the datatype sequence. However, the sequence datatype is not explicitly linked to the classes nucleotide and amino-acid defined in the ChEBI ontology5, recommended by OBO Foundry as a reference ontology.
In Fig. 2, we present the extension of OntoDT to represent the biosequence datatype from BioXSD. We represent the bio-sequence datatype class as a subclass of the character sequence datatype class with the defined semantic meaning in the NCI Thesaurus and the EDAM ontology6. In order to define the nucleotide and amino acid sequences datatypes, we define two subclasses of the character datatype class: nucleotide character datatype and amino acid character datatype. In order to define their semantic meaning, we explicitly link them to the nucleotide and amino acid classes from the ChEBI ontology. Consequently, the bio-sequence datatype class has two subclasses: nucleotide sequence datatype and amino 4 http://ncit.nci.nih.gov/ 5 http://www.ebi.ac.uk/chebi/ 6 http://edamontology.org/ acid sequence datatype. Furthermore, both datatypes have two subclasses, depending on whether they include ambiguous bases (in the case of nucleotides) or ambiguous and additional residues (in the case of amino acids). For example, the nucleotide sequence datatype class has two subclasses: nucleotide sequence with ambiguous bases (general nucleotide sequence in BioXSD) and nucleotide sequence without ambiguous bases (nucleotide sequence in BioXSD).
In a simmilar way, we represent the bio-sequence record datatype class as a subclass of the record datatype class. This datatype is defined by a record generator and the bio-sequence-field-list. As defined in BioXSD, the datatype contains a bio-sequence as a mandatory component and a set of metadata (such as name, note, species, translationalData, reference, inlineBaseQuality) as nonmandatory components. In OntoDT, we model the bio-sequence field component class is as a role of the bio-sequence datatype.
BioXSD uses a combined approach of a pure XML Schema
annotated by a data-type ontology using Semantic Annotations for
Web Services Description Language7 (WSDL) and XML Schema.
SAWSDL defines a set of extension attributes for the WSDL
and XML Schema definition languages. Application of attributes
allows the description of additional semantics by using references
to conceptual semantic models, e.g., ontologies. BioXSD datatypes
are annotated with terms from the EDAM ontology
The use case presented in this extended abstract demonstrates that OntoDT provides logically consistent representation of bioinformatics datatypes from BioXSD and enables an accurate representation of the semantic meanings of the data of specified datatypes. OntoDT has been designed as a generic and comprehensive ontology of datatypes and consequently any datatype from other resources can also be represented by OntoDT. We suggest that OntoDT can serve as a reference model for the consistent representation of datatypes used within biomedical domains and wider.