Connect via Zoom here!
Leveraging Infrared Imaging with Machine Learning for Phenotypic Profiling
Presented By: Xinwen Liu
Abstract:
Phenotypic profiling maps and analyzes phenotypes in biological systems under various conditions, aiding research in drug discovery, disease modeling and systems biology by understanding perturbation impacts. Current techniques face challenges such as high costs, complexity, and batch effects. Here, we developed a new phenotypic profiling method using infrared (IR) imaging coupled with machine learning, offering a high-throughput, cost-effective, and easily-operated alternative by capturing biochemical fingerprints of phenotypes. In particular, IR-active vibrational probes are systematically designed to enhance the metabolic specificity of IR imaging, improving sensitivity and specificity for phenotype discrimination. In addition, machine learning methods, including supervised learning, unsupervised learning and anomaly detection, are employed with additional algorithmic design for analyzing and mapping cell phenotypes under drug and disease perturbations. Overall, the developed method shows great promise in phenotypic drug discovery and disease modeling.
