SIB Young Bioinformatician Award 2012
Resolving the ortholog conjecture: orthologs tend to be weakly, but significantly, more similar in function than paralogs
Christophe Dessimoz, winner of the Young Bioinformatician Award, has several affiliations among which he is a visiting scientist at the European Bioinformatics Institute in the group of Dr. Nick Goldman, and a research associate in the SIB’s group of Prof. Gaston Gonnet at ETHZ.
Christophe lead a study that confirmed the long-held belief that studying the genes we share with other animals is useful. The study, published in the open access journal PLoS Computational Biology, showed how bioinformatics makes it possible to test the fundamental principles on which life science is built.
Studying genes helps life science researchers understand how our bodies work and how diseases progress. Scientists have long looked to model species – mice, for example – to understand human biology. This is at the root of what is called the ‘ortholog conjecture’: the idea that we can take what we learn from a few species and apply it to many.
For the past 40 years, scientists have thought that genes that directly descend from a common ancestor ("orthologs") tend to be functionally closer than imperfect copies of genes arising through gene duplication within the same species ("paralogs"), and this has worked quite well: studying genes in model species has provided invaluable insights in all areas of biology. But until now, there hasn’t been enough data to test this "ortholog conjecture" with authority. Now, with advances in biotechnology producing vast quantities of data every day, there is finally enough information on hand to settle the debate.
Using advanced computational techniques on data derived from tens of thousands of scientific articles, Dr. Dessimoz and his colleagues analysed 400 000 pairs of genes (orthologs and paralogs) from 13 different species. The team compared the two approaches and corroborated the ortholog conjecture.
SIB Best Graduate Paper Award 2012
Moment-Based Inference Predicts Bimodality in Transient Gene Expression
This year’s Best Graduate Paper co-winners are both working at the ETHZ (Swiss Federal Institute of Technology in Zurich). Jakob Ruess belongs to the Automatic Control Laboratory led by Prof. John Lygeros and Christophe Zechner belongs to the Biomolecular Signaling and Control Group led by Prof. Heinz Koeppl.
Jakob and Christoph developed a computational framework for calibrating stochastic gene expression models using experimental flow cytometry data. Their method allowed them to study the osmo-stress induced transcriptional response in budding yeast.
Recently, progress has been made in characterizing the impact of extrinsic variability on the dynamics of stochastic models. In their work, Jakob and Christoph addressed the corresponding inverse problem of inferring the intrinsic molecular- as well the extrinsic noise from experimental data. More specifically, they proposed a method to estimate kinetic parameters and to quantify the different sources of cell-to-cell variability. Using this method, they were able to build and calibrate a computational model for the expression of the stress-induced gene STL1. Their study provides additional support for several biological hypotheses, for instance that the experimentally observed bimodality in STL1 expression stems from the intrinsic randomness of the transcription initiation process.