Biomedical Engineering Reference
In-Depth Information
4 The Qualifier field can contain one or more of 'NOT', 'colocalizes_with' and
'contributes_to'. Of importance is the 'NOT' qualifier, which negates the
annotation.
5 The Evidence Code field describes the type of analysis or experiments on which the
annotation is based. The types of evidence code are listed at http://www
.geneontology.org/GO.evidence.shtml. The type of evidence code has an impact on the
credibility of the annotation. In particular, the code 'Inferred from Electronic
Annotation' (IEA) is assigned to annotations based on an automatic method without
curatorial judgement. The use of such annotations is not recommended and is often
avoided.
Notes
a
Function annotations for a protein may be found in more than one annotation file; for
example, the annotation files for Protein Data Bank, UniProt and possibly the organism's
specific database.
GO comprises three structured controlled vocabularies, or
ontologies
, for describing
molecular function, biological process and cellular component. Each ontology comprises
functional terms arranged in a hierarchical structure known as a
directed acyclic graph
(DAG). The DAG differs from a tree mainly in that a term in the former can have multiple
parent terms. The relationship between a parent and its children terms is also further specified
by two different relationships:
is_a
and
part_of
. Similar to the FunCat, each child term in
GO describes a more specific form of its parents. The current version of GO has 26 384
terms. Figure 9.2 illustrates the ancestor terms for 'nucleobase, nucleoside, nucleotide and
nucleic acid metabolic process' in the 'biological process' ontology.
Refer to Protocol 9.1 on how to obtain the GO scheme and annotation files. If you wish
to use an existing function prediction application for prediction, you may not need to obtain
the scheme and annotation data. Many applications are available as online Web services and
are already trained on existing annotation data. For such applications, only the features of
a protein, such as sequence or structure, are required for prediction. Some applications that
Figure 9.2
A subset of a functional category in GO (http://amigo.geneontology.org/cgi-bin/
amigo/term-details.cgi?term=GO:0006139). The term '
nucleobase
,
nucleoside
,
nucleotide
and
nucleic acid metabolic process
' has two parent terms '
cellular metabolic process
'and
primary
metabolic process
.'
