The de Boer Lab, whose current research is focused on understanding the logic and evolution of gene regulation, has just published a paper in Nature that offers an exciting new computer model, an “Oracle,” which was able to query vastly more DNA sequences than is possible in traditional experiments.
In the paper, the team (a collaboration with Dr. Aviv Regev’s and Francisco Cubillos’s groups, and first author Eeshit Dhaval Vaishnav) not only demonstrated the efficacy of the model but also produced findings that will help shape how researchers think about mutations in regulatory regions. “The whole paper is about evolution in yeast, so we’re not curing cancer just yet,” quips principal investigator, Dr. Carl de Boer, “but many of the same principles we’ve learned in yeast are likely to hold elsewhere, including in cancer.”
Yeast is a great model for evolution because researchers can easily mutate its genome and monitor how the mutations affect its ability to grow and reproduce. Like humans, yeast is a eukaryote, and gene regulation in both humans and yeast works in similar ways. “A lot of our findings in yeast are supported by similar data in humans,” says de Boer. But what’s even more promising in this paper is that this same computational ‘oracle’ they used to query DNA sequences was also useful for designing regulatory sequences. “This enabled us to create new sequences whose expression levels fell at—and beyond—the normal extremes of expression.” These have potential use cases in biotechnology in biosynthesizing chemicals with yeast.
Evolution is such a fascinating field! It’s why we’re all here, and also why we get cancer. What really drove this project is that we finally had data and computer models that were good enough to use as a gene regulation ‘oracle’.
– Carl de Boer
Work in the de Boer lab all stems from one of humankind’s most persistent questions distilled literally down to the genetic level: what are the mechanisms of evolution? As de Boer says, “Evolution is such a fascinating field! It’s why we’re all here, and also why we get cancer. What really drove this project is that we finally had data and computer models that were good enough to use as a gene regulation ‘oracle’. That enabled us to ask questions of the computer about how the functions of regulatory sequences change when you mutate them in different ways. From there, the project took off to explore evolution down exciting paths.”
It’s an important milestone for the de Boer lab as the next steps for this research is to deploy the method in new directions while improving the computational models to better predict gene expression. One of their key findings, for instance, is that there are often many ways for a sequence to evolve equivalent levels of expression. This has implications for what one expects to see in cancer mutations: namely, that seeing the same mutation over and over again across patients is not what we expect as tumours might all have taken different routes to get to the same point. The team even identified two yeast lineages that evolved the same regulatory program, but achieved this using different mutations.
The de Boer lab’s work is an exciting example of what researchers are discovering and building at the nexus of data science and biology. “Hopefully, this kind of thing spurs more computer scientists to get interested in the field of Biomedical Engineering,” says, de Boer. “We’re really just scratching the surface of what’s out there.”
This work was generously supported by: Canadian Institutes for Health Research Fellowship; NIH; ANID – Programa Iniciativa Científica Milenio; Klarman Cell Observatory
Dr. Carl de Boer Carl is the founding member of the lab. He did his PhD in the lab of Tim Hughes at the University of Toronto, and a post-doc at the Broad Institute in Aviv Regev’s group. He is currently an Assistant Professor in the School of Biomedical Engineering at UBC.
UBC School of Biomedical Engineering