IC Colloquium: Machine learning approaches to human disease
By: Michael Skinnider - University of British Columbia
IC/SV Faculty candidate
Abstract
Advances in genomic, transcriptomic, and proteomic technologies now enable the routine measurement of thousands of biomolecules within any biological sample. With the advent of single-cell techniques that multiplex these measurements over millions of cells, the scale of biological data generation is outpacing our human ability to comprehend this data. Approaches based on machine intelligence offer the possibility of augmenting our human capacity for understanding in order to enable biological discovery. In this seminar, I will explore applications of machine learning to enable biological discovery from genomic, proteomic, transcriptomic, and metabolomic datasets, and discuss implications for the diagnosis and treatment of human disease. Areas of focus will include the discovery of novel antibiotics from nature using genomic and metabolomic datasets; charting protein-protein interactions within mammalian tissues and across the tree of life using proteomic datasets; mapping neural circuits in high-throughput using single-cell and spatial transcriptomics datasets; and the identification of novel designer drugs of abuse from metabolomic datasets. Looking forward, I will explore how machine learning might bridge the most significant remaining gap in our ability to measure biological systems: namely, our ability to measure all of the small molecules within any biological sample. I will discuss the use of chemical artificial intelligence to identify unknown molecules within mass spectrometry data, and the implications of this technology for our understanding of cancer metabolism and germline cancer risk.
Bio
Michael Skinnider is an MD/PhD student at the University of British Columbia and a visiting PhD student at EPFL. Previously, he earned an Bachelor’s of Arts & Science from McMaster University. His undergraduate research, with Nathan Magarvey, leveraged bacterial genomes and metabolomes to discover new antibiotics from nature. Some of the technology he developed as an undergraduate is now being licensed by a successful spin-off company, Adapsyn Bioscience. His doctoral research, with Leonard Foster, focused on the use of proteomics to map cellular protein interaction networks. Separately, his work with Grégoire Courtine has employed single-cell and spatial transcriptomics to mechanistically dissect the recovery of walking after paralysis. More recently, he has developed chemical AI to automate the discovery of new small molecules, including human metabolites and designer drugs of abuse (with David Wishart and Petur Dalsgaard). He is originally from Victoria, British Columbia. His research interests are broadly at the intersection of applied machine learning for problems in biology, chemistry, and medicine, with a particular interest in the discovery of novel small molecules.
More information
IC/SV Faculty candidate
Abstract
Advances in genomic, transcriptomic, and proteomic technologies now enable the routine measurement of thousands of biomolecules within any biological sample. With the advent of single-cell techniques that multiplex these measurements over millions of cells, the scale of biological data generation is outpacing our human ability to comprehend this data. Approaches based on machine intelligence offer the possibility of augmenting our human capacity for understanding in order to enable biological discovery. In this seminar, I will explore applications of machine learning to enable biological discovery from genomic, proteomic, transcriptomic, and metabolomic datasets, and discuss implications for the diagnosis and treatment of human disease. Areas of focus will include the discovery of novel antibiotics from nature using genomic and metabolomic datasets; charting protein-protein interactions within mammalian tissues and across the tree of life using proteomic datasets; mapping neural circuits in high-throughput using single-cell and spatial transcriptomics datasets; and the identification of novel designer drugs of abuse from metabolomic datasets. Looking forward, I will explore how machine learning might bridge the most significant remaining gap in our ability to measure biological systems: namely, our ability to measure all of the small molecules within any biological sample. I will discuss the use of chemical artificial intelligence to identify unknown molecules within mass spectrometry data, and the implications of this technology for our understanding of cancer metabolism and germline cancer risk.
Bio
Michael Skinnider is an MD/PhD student at the University of British Columbia and a visiting PhD student at EPFL. Previously, he earned an Bachelor’s of Arts & Science from McMaster University. His undergraduate research, with Nathan Magarvey, leveraged bacterial genomes and metabolomes to discover new antibiotics from nature. Some of the technology he developed as an undergraduate is now being licensed by a successful spin-off company, Adapsyn Bioscience. His doctoral research, with Leonard Foster, focused on the use of proteomics to map cellular protein interaction networks. Separately, his work with Grégoire Courtine has employed single-cell and spatial transcriptomics to mechanistically dissect the recovery of walking after paralysis. More recently, he has developed chemical AI to automate the discovery of new small molecules, including human metabolites and designer drugs of abuse (with David Wishart and Petur Dalsgaard). He is originally from Victoria, British Columbia. His research interests are broadly at the intersection of applied machine learning for problems in biology, chemistry, and medicine, with a particular interest in the discovery of novel small molecules.
More information
Practical information
- General public
- Free
Contact
- Host: