Search papers, labs, and topics across Lattice.
This study systematically evaluates uncertainty quantification (UQ) methods in deep learning models applied to genomics, specifically comparing Deep Ensembles, Bayesian Neural Networks, and Monte Carlo-dropout techniques. The analysis reveals that Bayesian Neural Networks excel in capturing uncertainty related to class imbalance and out-of-distribution data, despite being computationally more intensive. Additionally, the research demonstrates the practical utility of uncertainty scores in enhancing the selection of high-quality predictions in protein-RNA interactions, providing valuable insights for future genomic applications.
Bayesian Neural Networks outperform other methods in handling uncertainty in genomics, revealing critical insights for high-stakes biological predictions.
Deep learning models have emerged as the standard computational tool for a wide range of applications in genomics. Yet, uncertainty quantification (UQ) -- and more specifically, the reliability of different uncertainty estimates in this domain -- has received little systematic attention. This work presents an empirical analysis of UQ in deep learning models, focusing on genomics applications. In a series of experiments, we contrast Deep Ensembles, Bayesian Neural Networks, and Monte Carlo-dropout methods. We assess their ability to quantify uncertainty in different scenarios, accounting for common dataset characteristics in two genomic application areas and modalities: sequence-to-activity models, and single-cell expression analysis. Our systematic comparison framework provides guidelines for the applicability and reliability of UQ methods in genomics, highlighting their strengths and limitations in different scenarios. We show that Bayesian Neural Networks are better at capturing uncertainty caused by strong class imbalance and out-of-distribution data in genomics, despite their computational disadvantages. Moreover, we show how uncertainty scores can be used to select high-quality predictions in protein-RNA interactions.
