Intraspecies variation in medaka fish and humans
Director of EMBL-EBI
EMBL-EBI Directors Office
EditThe Birney group applies novel AI methodologies to gain new insights into molecular, cellular and organismal biology, in particular methods which leverage genetic or genomic information.
We are interested in all aspects of using AI in molecular biology from fundamental genome sequence interpretation through to understanding the life course of individuals during health care. We use a variety of AI techniques, including generative pre-trained transformers, variational autoencoders, decoder-only architectures and convolution networks, and connect them to a variety of data inputs, such as large scale health care records, organ-scale imaging, single cell data and protein or DNA sequence information. We often bring in genetic information both as potential additional data items or via the loss function. We are expert users in DNA sequence interpretation, including Oxford Nanopore Technology, which can provide more information via the length of reads or the modification status of the nucleotides. We can use any biological system, often human but also Medaka fish, Zebra fish, Mouse, C. elegans or Drosophila. We are interested in both expanding the methodological reach of AI for new tasks and new data items, and in using AI to gain new insights into biology.
Figure showing exploration of human heart physiology. Panel A shows a dissected human heart open on the left ventricle, showing the complex rough surface facing the blood pool. Panel B shows an idealised heart and the 2D MRI scans done in the UK BioBank. Panel C shows the results of a deep learning technique to find the muscle wall and then the follow-on derivation for the fractal dimension of the ventricle. This was later used for human genetic investigation, including causal analysis of the role of trabeculation in heart function.