Yes, Ecology can improve Genomics

Blogging on Peer-Reviewed ResearchFew organisms are as well understood at the genetic level as Saccharomyces cerevisiae. Given that there are more yeast geneticists than yeast genes and exemplary resources for the community (largely a result of their size), this comes as no surprise. What is curious is the large number of yeast genes for which we’ve been unable to characterize. Of the ~6000 genes currently identified in the yeast genome, 1253 have no verified function (for the uninclined, this is roughly 21% of the yeast proteome). Egads! If we can’t figure this out in yeast, what hope do we have in non-model organisms?Lourdes Peña-Castillo and Timothy R. Hughes discuss this curious observation and its cause in their report in Genetics.

The authors point out several interesting things about these genes of unknown function that throw immediate skepticism out the window: most were identified in the original genome annotation (so they aren’t “new” discoveries without a chance to be analyzed), most show signs of conservation with homologs across taxa and evidence of expression, (suggesting that they are real genes), and these genes are routinely identified in large scale analyses (pulldowns, expression profiling, etc.), indicating that they are not being missed by researchers.

Finally, the authors explore the possibility that redundant function annotation classification when these genes are knocked out. While 10% of the unclassified genes appear to be paralogs, this hypothesis cannot explain away the other 90%.It may be, they go on to propose, that researchers are looking at yeast in the wrong conditions. Knock out a gene vital to warding off an amoeba attack in nature and you’ll only observe its phenotype in the presence of an attacking amoeba. Here the authors expose an avenue of yeast biology that has been largely relatively ignored: yeast ecology. If we understood more about how yeast behaves in nature, we could better design experiments to test the functions of these unknown genes in the lab. By parameterizing yeast as a bug that grows on a petri dish, we inherently limit our consideration of how yeast evolved and, hence, what it can do.

This isn’t meant to be a criticism of the yeast community. What they’ve done in the ten years since the yeast genome was released is amazing and should serve to remind us of just what can be done when a community comes together to solve large problems. Rather, I see this as a call to genetics and genomics researchers of any organism. To understand a gene, we should consider, in addition to the genetic, biochemical and cellular environment, the gene’s evolutionary trajectory and its role in the organism’s interaction with the natural environment. Lots of attention has been placed on obtaining sequence, but unless we begin to focus on ecology, there may be a large knowledge gap that prevents us from fully understanding all of the sequence data we’ve gathered. Deconstructistic thought has dominated a great deal of contemporary biology; view an organism as a bag of genes and survey each gene, one at a time, to understand the organism. This has been a largely successful approach, but can only take us so far. It would appear, based on the synopsis by Peña-Castillo and Hughes, that the wall has been hit in yeast genetics. To get over it, we may, as a community, need to adopt a more synthetic based approach to problem solving, borrowing ideas from many fields, especially ecology, to solve single problems. It wouldn’t surprise me if we see Saccharomyces cerevisiae, as a result of the proactive nature of its research community members, become a model ecological organism in the near future.

One thought on “Yes, Ecology can improve Genomics”

  1. An even simpler problem to solve would be to get scientists that are doing functional characterization to trust more the computational predictions and to try and follow them. I suspect that most groups work from their main focus proteins outwards by fishing for new potential associated proteins. It looks like a group could make a lot of interesting findings by aiming directly at the list of unknowns and using functional predictions to guide the experiments.

Leave a Reply