Biosketch

Mafalda Dias:

Jonathan Frazer:

Summary

We are entering an era of population scale sequencing of humans, global efforts to obtain reference genomes for all life on Earth, and experiments that can assay the effects of millions of genetic variants. These datasets contain the information to transform our use of genomic data in diagnosis and preventative care, in protein and drug design, and much more, but we need new computational strategies to extract this information. We develop machine learning methods (often generative, often Bayesian) to predict the effect of genetic variation on phenotype, with an emphasis on research which will directly impact the diagnostic yield of patient sequencing.

From a machine learning perspective, we are interested in how recent developments in deep learning may be adapted for modelling genetic sequence data, which poses modelling challenges unlike other data typically seen in machine learning. From an evolutionary biology perspective, we are interested in the relationship between fitness and phylogeny and how the genetic variation seen on different evolutionary timescales, from within populations, to across the entire tree of life, can be used to learn about disease and molecular function.

Job Openings

We are building a team, and hiring at all levels. If interested, please get in contact with Mafalda and Jonathan.