- Annotation of genes with GO terms can be represented as a bipartite network,
with gene and GO term sets representing partitions and edges between partitions
representing gene annotations. - The hierarchical structure of GO is represented as a directed acyclic graph
(DAG), where nodes represent biological terms (GO terms) and directed edges
represent parent-child semantic relations.
GO data (ontology and annotation data) can be obtained from GO data-
base (Table 1 ). When constructing the DAG, we usually use all possible semantic
relations between GO terms:is_a,regulates,has_part, andpart_of.
Human disease associations are obtained from Disease Ontology (DO). DO
provides the biomedical community with a consistent and unique ontological
disease classification by extensive cross mapping of DO terms to well-established
and well-adopted terminologies of MeSH,^88 ICD,^89 NCIs thesaurus,^90 SNOMED,^91
and OMIM.^92 The semantic structure of DO can also be represented as a DAG,
where DO terms are linked byis_arelations in the hierarchy.
3.2 Network Representations
A usual way to store biological data describing molecular networks is by using
matrix representation. Such a representation provides an easy way of manipulation
and processing of the data in many data integration algorithms. Here, we introduce
basic concepts from graph theory that are often used in data integration methodol-
ogies and explain them on a simple example shown in Fig. 2.
To represent graphs, we often use the concept of anadjacency matrix.An
example of a graph containing 6 nodes is represented in Fig. 2 (B), and its
corresponding adjacency matrix is represented in Fig. 2 (C). An adjacency matrix,
A, of a graphG(V,E) is a square matrix of dimension |V||V|,^93 where each
column and row correspond to a node; an element of an adjacency matrix of a
un-weightedgraph is ether 0 for nonadjacent nodes or 1 for adjacent nodes. In case
ofweightedgraphs, nonzero entries in an adjacency matrix correspond to link
(^88) Nelson et al. ( 2004 ).
(^89) Ayme et al. ( 2010 ).
(^90) Sioutos et al. ( 2007 ).
(^91) Cornet and de Keizer ( 2008 ).
(^92) Hamosh et al. ( 2005 ).
(^93) |V| stands for the number of elements in set V.
154 V. Gligorijevic ́and N. Pržulj