1Protein Data Integration through Ontologies
Amandeep S. Sidhu
Digital Ecosystems and Business Intelligence Institute,
Curtin University of Technology, Perth, Australia
a.sidhu@curtin.edu.au
Abstract. In this paper, we consider the challenges of information integration in
proteomics from the prospective of researchers using information technology as
an integral part of their discovery process. Specifically, data integration, meta-
data specification, data provenance and data quality are discussed here. Here we
review existing protein data integration methods and propose the use of common
vocabulary for protein data integration using ontologies.
1. Introduction
The advent of automated and high-throughput technologies in biological research and
the progress in the genome projects has led to an ever-increasing rate of data
acquisition and exponential growth of data volume. However, the most striking
feature of data in life science is not its volume but its diversity and variability. The
biological data sets are intrinsically complex and are organised in loose hierarchies
that reflect our understanding of complex living systems, ranging from genes and
proteins, to protein-protein interactions, biochemical pathways and regulatory
networks, to cells and tissues, organisms and populations, and finally ecosystems on
earth. This system spans many orders of magnitudes in time and space and poses
challenges in informatics, modelling, and simulation that goes beyond any scientific
endeavour. Reflecting the complexity of biological systems, the types of biological
data are highly diverse. They range from plain text of laboratory records and literature
publications, nucleic acid and protein sequences, three-dimensional atomic structure
of molecules, and biomedical images with different levels of resolutions, to various
experimental outputs from technology as diverse as microarray chips, light and
electronic microscopy, Nuclear Magnetic Resonance (NMR), and mass spectrometry.
This presents a great challenge in modelling biological objects. In this paper we will
discuss existing protein data integration methods and propose the use of common
vocabulary for protein data integration using ontologies.
2. Related Works
2.1 Existing Data Integration Methodologies
In the context of protein data, annotation generally refers to all information about
protein other than protein sequence. Traditional approaches to integrate protein data
generally involved keyword searches, which immediately excludes unannotated or
poorly annotated data. It also excludes proteins annotated with synonyms unknown to
the user. Of the protein data that is retrieved in this manner, some biological resources
do not record information about the data source, so there is no evidence of the
Na
tu
re
P
re
ce
di
ng
s :
h
dl:
10
10
1/
np
re
.2
00
9.
40
62
.1
:
Po
ste
d
9
De
c 2
00
9

2annotation. An alternative protein annotation approach is to rely on sequence identity,
or structural similarity, or functional identification. The success of this method is
dependent on the family the protein belongs to. Some proteins have high degree of
sequence identity, or structural similarity, or similarity in functions that are unique to
members of that family alone. Consequently, this approach can’t be generalised to
integrate the protein data. Clearly, these traditional approaches have limitations in
capturing and integrating data for Protein Annotation. Perhaps these problems could
be addressed more easily in the context of a more general logical structure. For these
reasons, we have adopted an alternative method that does not rely on keywords or
similarity metrics, but instead uses ontology. Ontology is a means of formalising
knowledge; at the minimum ontology must include concepts or terms relevant to the
domain, definitions of concepts, and defined relationships between the concepts.

2.2 Need for Biomedical Ontologies
Semantics of protein data is usually hard to define precisely because they are not
explicitly stated but are implicitly included in database design. Proteomics is not a
single, consistent domain; it is composed of various smaller focused research
communities, each having a different data format. Data Semantics would not be a
significant issue if researchers only accessed data from within a single research
domain, but this is not usually the case. Typically, researchers require integrated
access to data from multiple domains, which requires resolving terms that have
slightly different meanings across communities.
To integrate the data generated through a web-based system the research
results need to be consistent, classified, retrieved and queried using a unified common
vocabulary. This will facilitate sharing and cross-linkage of the results and will
provide mechanism for interoperability between various databases. We note most of
the modern biological databases use data descriptors specified in a schema curated
according to the needs and requirements of the immediate community, without
consideration to interoperability with other databases. This underlying issue of
heterogeneity in biological domain can be addressed partly by developing a common
vocabulary using Ontologies for data modelling and knowledge sharing as
demonstrated in Genomics by Gene Ontology (Lewis, 2004, Ashburner et al., 2001)
and MGED Ontology (Whetzel et al., 2006).
The Gene Ontology is a collaborative effort to create a controlled vocabulary
of gene and protein roles in cells, addressing the need for consistent descriptions of
gene products in different databases. The GO collaborators are developing three
structured, controlled vocabularies (ontologies) that describe gene products in terms
of their associated biological processes, cellular components and molecular functions
in a species-independent manner. One of the important uses of GO is the prediction of
gene function based on patterns of annotation. For example, if annotations for two
attributes tend to occur together in the database, then the gene holding one attribute is
likely to hold for the other as well (King et al., 2003). In this way, functional
predictions can be made by applying prior knowledge to infer the function of the new
entity (either a gene or a protein).
Na
tu
re
P
re
ce
di
ng
s :
h
dl:
10
10
1/
np
re
.2
00
9.
40
62
.1
:
Po
ste
d
9
De
c 2
00
9

3 The MGED Ontology (MO) is developed by the Microarray Gene Expression
Data (MGED) Society. MO provides terms for annotating all aspects of a microarray
experiment from the design of the experiment and array layout, through to preparation
of the biological sample and protocols used to hybridise the RNA and analyse the
data. MO is a species-neutral ontology that focuses on commonalities among
experiments rather than differences between them. MO is primarily an ontology used
to annotate microarray experiments; however, it contains concepts that are universal
to other types of functional genomics experiments. The major component of the
ontology involves biological descriptors relating to samples or their processing.
3. Protein Ontology (PO)
We built the Protein Ontology (Sidhu et al., 2007; Sidhu et al., 2005; Sidhu et al.,
2004) to integrate protein data formats and provide a structured and unified
vocabulary to represent protein synthesis concepts. Following PO's lead recently two
other ontologies have been designed for as well. PRotein Ontology (PRO) (Natale et
al., 2007) to facilitate protein annotation and to guide new experiments. The
components of PRO extend from the classification of proteins on the basis of
evolutionary relationships to the representation of the multiple protein forms of a
gene. Proteomics Process Ontology (ProPreO) (Sahoo et al., 2006) enables a detailed
description of proteomics experimental processes and data.
Protein Ontology (PO) provides an integration of heterogeneous protein and
biological data sources. It converts the enormous amounts of data collected by
geneticists and molecular biologists into information that scientists, physicians and
other health care professionals. PO consists of concepts, which are data descriptors
for proteomics data and the relationships among these concepts. PO has (1) a
hierarchical classification of concepts, from general to specific; (2) a list of properties
related to each concept; (3) a set of relationships to link concepts in ontology in more
complicated ways then implied by the hierarchy; and (4) a set of algebraic operators
for querying protein ontology instances. In this section, we will briefly discuss
various concepts and relationships that make up PO. More details about Protein
Ontology are available on the website (http://proteinontology.org.au/).
3.1 Protein Ontology Concepts
The root concept in PO is ProteinOntology. For each instance of protein that is
entered into PO, the submission information is entered for ProteinOntology concept.
There are seven concepts of PO, called Generic Concepts that are used to define
complex PO Concepts: {Residues, Chains, Atoms, Family, AtomicBind, Bind, and
SiteGroup}. These generic concepts are reused in defining complex PO concepts. We
now briefly describe these generic concepts. Details and Properties of Residues in a
Protein Sequence are defined by instances of Residues concept. Instances of Chains of
Residues are defined in Chains concept. All the Three Dimensional Structure Data of
Protein Atoms are represented as instances of Atoms concept. Defining Chains,
Residues and Atoms as individual concepts has the advantage that any special
properties or changes affecting a particular chain, residue and atom can be easily
Na
tu
re
P
re
ce
di
ng
s :
h
dl:
10
10
1/
np
re
.2
00
9.
40
62
.1
:
Po
ste
d
9
De
c 2
00
9

4added. Family concept represents Protein Super Family and Family Details of
Proteins. Data about binding atoms in Chemical Bonds like Hydrogen Bond, Residue
Links, and Salt Bridges are entered into ontology as an instance of AtomicBind
concept. Similarly, the data about binding residues in Chemical Bonds like
Disulphide Bonds and CIS Peptides is entered into ontology as an instance of Bind
concept. When defining the generic concepts of AtomicBind and Bind in PO we again
reuse the generic concepts of Chain, Residue, and Atom. All data related to site groups
of the active binding sites of Proteins are defined as instances of SiteGroup concept.
In PO, the notions classification, reasoning, and consistency are applied by defining
new concepts from the defined generic concepts.
The Main Concept for definition of Protein Complexes in the Protein
Ontology is ProteinComplex. ProteinComplex concept defines one or more Proteins
in the Complex Molecule. Six sub concepts of ProteinComplex: Entry, Structure,
StructuralDomains, FunctionalDomains, ChemicalBonds, and Constraints provide a
complete understanding of the sequence, structure and functional interactions of
proteins. They define sequence, structure, function, and chemical bindings of the
Protein Complex and are derived concepts formed from the generic concepts
discussed earlier.
PO describes Protein Complex Entry and the Molecules contained in Protein
Complex are described using Entry concept and its sub-concepts of Description,
Molecule and Reference. Molecule reuses the generic concept of Chain to represent
the linkage of molecules in the protein complex to the chain of residue sequences.
Protein Sequence and Structure data are described using Structure concept in
PO with sub-concepts ATOMSequence and UnitCell. ATOMSequence represents
protein sequence and structure and is made of generic concepts of Chain, Residue and
Atom. Protein Crystallography Data is described using the UnitCell concept.
Protein Structural Folds and Domains are defined in PO using the derived
concept of StructuralDomains. Family and Super Family of the organism in which
protein is present are represented in StructuralDomains by reference to the generic
concept of Family. Structural Folds in protein are represented by sub-concepts of
Helices, Sheets and Other Folds. Each definition of structural folds and domains also
reuses the generic concepts of Chain and Residue for describing the Secondary
Structure of Proteins.
PO has the first Functional Domain Classification Model for proteins defined
using the derived concept of FunctionalDomains. Like StructuralDomains, the
Family and Super Family of the organism in which protein is present, are represented
in FunctionalDomains by reference to the generic concept of Family.
FunctionalDomains describes the Cellular and Organism Source of Protein using
SourceCell sub-concept, Biological Functionality of Protein using BiologicalFunction
sub-concept, and describes Active Binding Sites in Protein using ActiveBindingSites
sub-concept. Active Binding Sites are represented in PO as a collection of various Site
Groups, defined using SiteGroup generic concept.
Various chemical bonds used to bind various substructures in a complex
protein structure are defined using ChemicalBonds concept in PO. Chemical Bonds
are defined by their respective sub-concepts are: DisulphideBond, CISPeptide,
Na
tu
re
P
re
ce
di
ng
s :
h
dl:
10
10
1/
np
re
.2
00
9.
40
62
.1
:
Po
ste
d
9
De
c 2
00
9

5HydrogenBond, ResidueLink, and SaltBridge. They are defined using generic
concepts of Bind and Atomic Bind.
Various constraints that affect the final protein structural conformation are
defined using the Constraints concept of PO. The constraints described in PO at the
moment are: Monogenetic and Polygenetic defects present in genes that are present in
molecules making proteins described using GeneticDefects sub-concept, Hydrophobic
properties of proteins described using Hydrophobicity sub-concept, and Modification
in Residue Sequences are described using in ModifiedResidue sub-concept.
3.2 Relationships Protein Ontology
Semantics in protein data is normally not interpreted by annotating systems, since
they are not aware of the specific structural, chemical and cellular interactions of
protein complexes. A Protein Ontology Framework provides a specific set of rules to
cover these application specific semantics. The rules use only the relationships whose
semantics are predefined in PO to establish correspondence among terms. The set of
relationships with predefined semantics is: {SubClassOf, PartOf, AttributeOf,
InstanceOf, and ValueOf}. The PO conceptual modelling encourages the use of
strictly typed relations with precisely defined semantics. Some of these relationships
(like SubClassOf, InstanceOf) are somewhat similar to those in RDF Schema (W3C-
RDFSchema 2004) but the set of relationships that have defined semantics in our
conceptual PO model is too small to maintain the simplicity of the model. The
following is a brief description of the set of pre-defined semantic relationships in our
common PO conceptual model. SubClassOf relationship is used to indicate that one
concept is a specialisation of another concept. AttributeOf relationship indicates that a
concept is an attribute of another concept. PartOf relationship indicates that a concept
is a part of another concept. InstanceOf relationship indicates that an object is an
instance of the concept. ValueOf relationship is used to indicate the value of an
attribute of an object. By themselves, the relationships described above do not impose
order among the children of the node. We defined a special relationship called
Sequence(s) in PO to describe and impose order in complex concepts defining
Structure, Structural Folds and Domains and Chemical Bonds of Proteins.
Na
tu
re
P
re
ce
di
ng
s :
h
dl:
10
10
1/
np
re
.2
00
9.
40
62
.1
:
Po
ste
d
9
De
c 2
00
9

6™⌤┦✨?
✪┫Ⱖ∭✮⴬⬯⬣〧ㄢ☥㈥㌴〧㔯㈶㘢㜢㌲Ⱒ⬭
?
?
Na
tu
re
P
re
ce
di
ng
s :
h
dl:
10
10
1/
np
re
.2
00
9.
40
62
.1
:
Po
ste
d
9
De
c 2
00
9

74. SUMMARY
Nowadays many computational systems and databases have been developed to
provide analysis and information on proteins. However, integration of the information
is needed and so far this has not been possible as there was no common vocabulary
available that could be used as a standard language. Protein Ontology is a standard for
representing protein data in a way that helps in defining data integration and data
mining models for protein structure and function. It provides a unified controlled
vocabulary both for annotation data types and for annotation data. It is accepted as
part of Standardized Biomedical Ontologies at the National Center for Biomedical
Ontologies (http://bioportal.bioontology.org/ontologies/3905) along with Gene
Ontology and other biomedical ontologies.
REFERENCES
ASHBURNER, M., BALL, C. A., BLAKE, J. A., BUTLER, H., CHERRY, J. C., CORRADI, J. &
DOLINSKI, K. (2001) Creating the Gene Ontology Resource: Design and Implementation.
Genome Research, 11, 1425-1433 [PMID: 11483584].
KING, O. D., FOULGER, R. E., DWIGHT, S., WHITE, J. & ROTH, F. P. (2003) Predicting gene
function from patterns of annotation. Genome Research, 13, 896-904 [PMID: 12695322].
LEWIS, S. E. (2004) Gene Ontology: looking backwards and forwards. Genome Biology, 6,
103.1-103.4 [PMID: 15642104].
NATALE DA, ARIGHI CN, BARKER WC, BLAKE J, CHANG TC, HU Z, LIU H, SMITH B, WU
CH. Framework for a protein ontology. BMC Bioinformatics. 2007, 8 Suppl 9:S1 [PMID:
18047702].
SAHOO SS, THOMAS C, SHETH A, YORK WS, TARTIR S: Knowledge modeling and its
application in life sciences: a tale of two ontologies. In Proceedings of the 15th International
Conference on World Wide Web Edited by: Carr L, De Roure D, Iyengar A, Goble CA, Dahlin
M. New York, ACM Press; 2006, 317-326.
SIDHU, A. S., DILLON, T. S. & CHANG, E. (2007) Protein Ontology. IN CHEN, J. & SIDHU, A. S.
(Eds.) Biological Database Modeling. New York, Artech House.
SIDHU, A. S., DILLON, T. S. & CHANG, E. (2005) An Ontology for Protein Data Models. In Zhang,
Y.T., Roux, C. and Zhuang, T.G. (eds), 27th Annual International Conference of the IEEE
Engineering in Medicine and Biology Society (EMBC 2005). IEEE Engineering in Medicine
and Biology Society, Shanghai, 6120-6123 [PMID: 17281660].
SIDHU, A. S., DILLON, T. S., SIDHU, B. S. & SETIAWAN, H. (2004) A Unified Representation of
Protein Structure Databases. IN REDDY, M. S. & KHANNA, S. (Eds.) Biotechnological
Approaches for Sustainable Development. India, Allied Publishers.
WHETZEL, P. L., PARKINSON, H., CAUSTON, H. C., FAN, L., FOSTEL, J., FRAGOSO, G.,
GAME, L., HEISKANEN, M., MORRISON, N., ROCCA-SERRA, P., SANSONE, S.,
TAYLOR, C., WHITE, J. & STOECKERT, C. J. (2006) The MGED Ontology: a resource for
semantics-based description of microarray experiments. Bioinformatics, 22, 866-873 [PMID:
16428806].
Na
tu
re
P
re
ce
di
ng
s :
h
dl:
10
10
1/
np
re
.2
00
9.
40
62
.1
:
Po
ste
d
9
De
c 2
00
9