Deep learning expert (Postdoc or similar)

Application deadline:

"The newly established Single Cell and Synthetic Genomics lab at CRG Barcelona is looking for a deep learning expert (at postdoc level or similar), funded for 2 years, to join us in our effort to re-write the code of life. Two decades after the human genome was first sequenced, our ability to read and interpret the genetic code is still limited. While we have a good understanding of the ‘hardware’ of the genome (that is, the genes), our understanding of the ‘software’ that controls these genes, so called gene regulatory elements, lags behind. Importantly, these elements differ between individuals much more than the actual genes, determining our physical appearance, susceptibility to disease, and even our willingness to engage in extreme sports. A better understanding of this genomic ‘software’ therefore is of high interest to the pharmaceutical and biotechnological industry. Deep learning is a promising new tool to interpret DNA sequences. However, deep learning is very data hungry, and the human genome is of limited size: The genome may just be too small to truly learn to decipher these highly combinatorial rules. Building on a set of technologies that we have developed (Velten et al., 2017, Schraivogel et al., 2020) our lab therefore takes a different approach: We synthesize 10,000s of new regulatory elements, thereby systematically exploring parts of the very complex space of possibilities. By placing these new pieces of ‘software’ into different ‘environments’ (that is, different types of cells) we collect millions of data points, a rich resource to train deep neural nets. Our synthetic approach allows us to consistently create the type of data most informative to the model. Ultimately, it will not only allow us to interpret the genomic code, but also to write completely new, meaningful pieces of DNA. These can, for example, be used to design viruses that kill cancer cells, but not normal cells. Your role will be to contribute experience with deep learning to our team of bioinformaticians and experimental biologists. "