Concepts of Similarity in Bioinformatics - Essays in Bioinformatics - page 13

Recurrence

External
(in space & time)

Internal

Entities Relationships

Pattern Symmetry Harmony,

proportions

Figure 1. Simplified overview of concepts underlying structural descriptions.

Descriptions rely on elements (entities) and binary relationships between them [1,
13] (Table 1).
In the case of molecules, both the elements (substructures) and the relationships can
be described in terms of systems of categories. The categories and the relations between
them can be formalized into ontologies, which include the definitions of the elements as
well as the operations that are possible within the system ( Figure 2 ). Ontologies give
itemized descriptions each functions and roles a molecule can fulfil, so it is a logically
coherent world description. Entity-relationship-descriptions are generally applicable and
can be extended to such concepts as similarity groups, vicinities and networks ( Figure 2 ).

Table 1. Examples of models and descriptions

System Entities Relationships

a) Conceptual models of natural systems
Molecules Atoms Atomic interactions

(chemical bonds)

Assemblies Proteins, DNA Molecular contacts

Pathways Enzymes Chemical reactions

(substrates/products)

Genetic networks Genes Co-regulation

b) Structural descriptions
Protein structure

Atoms Chemical bonds

Protein structure Secondary structures Sequential and

topological vicinity

Folds Peptide bond

C D atoms

Protein sequence Amino acid Sequential vicinity

Elements and relationships can be described not only in terms of categories, but we
can assign to them property descriptors, such as physicochemical, chemical descriptors. In
terms of contents, there are two kinds of properties in proteins and DNA that deserve
special attention. i) The position of an element (nucleotide, atom) can be defined either
within the molecular chain (sequential position, with respect to the N-terminus, etc.) or in
as 3-D coordinates. ii) The function is a property or role that can be defined in the context
of a higher level. E.g. “protease” is a function defined either in an in vitro (e.g. action on a
certain substrate) or in vivo environment (e.g. role in complement activation). In addition to
these two main classes, there are a whole list of properties that can be assigned to entities