Machine Learning for Biodiversity Informatics - Lila Kari

Abstract: Even though biologists discover and classify thousands of new species every year, it is estimated that 95% of the over 20 million multicellular species on Earth do not yet have a scientific name or classification. The long term objectives of our research tie in with the Planetary Biodiversity Mission (map all multicellular life on Earth by 2045), and with deciphering the "Rosetta Stone'' of genomics (understand the semantics and utility of the mathematical structure of genomic sequences).
In this talk, Lira will discuss several mathematical representations of DNA sequences, and their use in conjunction with supervised machine learning, and unsupervised deep learning techniques for ultrafast, accurate, and scalable genome classification at all taxonomic levels. She will present recent findings of comprehensive evidence suggesting that adaptations to extreme temperatures and pH imprint a discernible environmental component in the genomic signature of microbial extremophiles. In particular, the unsupervised learning of unlabelled sequences identified several exemplars of hyperthermophilic organisms with large similarities in their genomic signatures, in spite of belonging to different domains in the Tree of Life.
Bio: Lila Kari is Professor in the School of Computer Science at the University of Waterloo, where she moved from her previously held position in the Department of Computer Science at the University of Western Ontario (1993-2017). She received her M.Sc. in 1987 from the University of Bucharest, Romania, and her Ph.D. in 1991. Author of more than 250 peer reviewed articles, Professor Kari is a recognized expert in the area of biomolecular computation, that is using biological, chemical and other natural systems to perform computations. She was the University Research Chair from 2017 to 2023. Her current research focuses on comparative genomics, biodiversity informatics, as well as theoretical aspects of bioinformation and biocomputation.