Martijn A. Huynen

Nijmegen Center for Molecular Life Sciences. Center for Molecular and Biomolecular Informatics, Toernooiveld 1, 6525 ED Nijmegen, Netherlands.
huynen@cmbi.kun.nl

The growing number of completely sequenced genomes and other types of genomics data provide invaluable information for protein function prediction. Compared to homology-based function prediction that provides information about the molecular function of a protein, the analysis of the gene in its genomic context or its co-expression with other genes provides information about its functional context: i.e. in which biological process does the protein play a role. The caveat of using the latter type of information is that it is very noisy: the co-expression of two genes is only a very weak indication that their proteins functionally interact with each other. One way to increase the reliability of such noisy functional genomics data is to include evolutionary conservation. As is the case for gene-order conservation in prokaryotes, conservation of co-expression between species dramatically increases the likelihood that proteins interact. The same can be observed when comparing yeast-2-hybrid data from multiple species. Interestingly, not only conservation of co-expression of genes between species, but also with a species, after parallel gene duplication, increases the likelihood of functional interaction. I will discuss these methods, highlighting them with several examples of protein function prediction based on the combination and integration of genomics data.