The example consists in a graph of 16 proteins and 19 interactions, from
Saccharomyces cerevisiae and/or from
Homo sapiens. These proteins and interactions have been annotated with NCBI taxonomic identifiers of this two species (4932 and 9606 respectively). To make such an annotation, we have used meta-genes obtained by Stuart et al. and BIND (see references) for the interactions part. Meta-genes are an abstraction that defines a canonical form of the ancestor of two orthologous genes (in this case, orthologs are from
Saccharomyces cerevisiae and
Homo sapiens).
In the obtained graph, an edge is drawn between two nodes if there is an interaction between the two proteins in at least one specie. This graph contains meta-genes, proteins from
Saccharomyces cerevisiae and a virtual protein complex (see
User manual for more information on virtual complex).
Using the example
Using the filter on taxonomic identifiers, we can obtain the three following graphs :
fig 1 : general graph
fig 2 : general graph filtered with Saccharomyces cerevisiae taxonomic identifier
fig 3 : general graph filtered with Homo Sapiens taxonomic identifier
We can see on the figure 2 that all proteins and interactions of the general graph are stil in this filtered view. Considering the designing strategies and the amount of interaction available for the yeast compared to those available in human, this is not surprising.
A more interesting analysis can be made on the third figure. The first trivial observation is that yeast proteins have desappeared. We can also see that there is only one remaining interaction after the filter, thus indicating two hypothesis : the interaction might have desappeared, and orthologs of yeast proteins are no more in interaction in human; the other hypothesis is that there are too few results in human interactome characterization to have found the same interactions that yeast ones.
This observation may indicate to the biologist that it would probably be a good idea to search interactions between these human proteins, since their orthologs in yeast are in interaction. To increase this hypothesis, we can then check annotations on the proteins (either using controled vocabularies or manual annotation filters).
Conclusion
This shot example as been designed to highlight that, using ProViz to explore a graph of interactions, we can create different views of the same graph using arbitrary properties, to reveal interesting informations. The example of orthologous genes is a nice one, but it can be extended to other domains. As an example, some organisms have reference databases that give the list of existing genes/proteins. So, a user may want to filter this graph and only keep proteins/genes that are in the list given by the reference database. To do this, and aknowledging that his graph file contains corresponding informations, he only have to filter proteins that have an xref to this database, which is pretty simple.