Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about a comprehensive benchmarking study that evaluates computational methods for predicting cellular gene expression changes following perturbations in this 57-minute conference talk. Discover how researchers created a systematic evaluation platform combining 11 large-scale perturbation datasets with expression forecasting software to assess various machine learning approaches used in drug discovery and cell fate engineering. Explore the key findings that performance heavily depends on metric selection, and understand why many sophisticated expression forecasting methods surprisingly fail to outperform simple baseline approaches when evaluated using basic metrics like mean squared error. Examine the implications for developmental genetics, cell fate engineering, and drug discovery applications, while gaining insights into the current limitations and future directions for improving computational expression prediction methods in biological systems.
