Biology Reference
In-Depth Information
Our method is implemented in Homolonto (Parmentier and Robinson-
Rechavi, unpublished), a software that we have developed in Java.
Compared to manual alignment of the ontologies, Homolonto reduces
time considerably, with high sensitivity. Thus, aligning the zebrafish
(ZFIN; 2087 terms) and Xenopus (Xenbase; 480 terms) ontologies took
1 month by hand, but 2 days using Homolonto. The first 213 pairs pro-
posed to the expert (i.e. 1 day of work) were valid at 80%, and contained
91% of all true positives.
3.3. Bgee,a Database for Gene Expression Evolution
To be useful for evo-devo and other comparative studies, the homology
relationships determined using Homolonto are implemented in a data-
base, Bgee. This “dataBase for Gene Expression Evolution” is being
developed to facilitate comparisons of gene expression between animal
species (http://bgee.unil.ch).
83
To enable the comparison of large-scale gene expression patterns,
Bgee must answer three conditions:
(a)
precise description of the anatomy and developmental stages of each
species, stored in a computer-understandable way. This is done
using existing ontologies, such as ZFIN
82
;
(b)
comparison criteria between anatomies, developmental stages, and
genes. For anatomy, this is done using Homolonto. For develop-
ment, we have developed a small ontology of “metastages” that are
common to all bilaterian animals, such as “blastula part_of embryo”.
For genes, we use homology predictions from other sources
(i.e. Ensembl
84
); and
(c)
integration of expression data in order to know in which anatomical fea-
tures (spatial mapping) and at which developmental stages (temporal
mapping) genes are expressed.
The relationships between these types of information are represented in
a very simplified manner in Fig. 4.
Concerning developmental stages, we came to the conclusion that it
is not possible to detail homologous stages in a similar way to organs.
