|
Abstract:
|
The location, timing, and abundance of gene expression within a tissue define the molecular mechanisms of cell functions. Recent technology breakthroughs in spatial molecular profiling, including imaging-based technologies and sequencing-based technologies, have enabled the comprehensive molecular characterization of single cells while preserving their spatial and morphological contexts. This new bioinformatics scenario calls for effective and robust computational methods to identify genes with spatial patterns. We represent two novel Bayesian hierarchical models to analyze spatial molecular profiling data, with several unique characteristics. The first model based on Gaussian process directly models the zero-inflated and over-dispersed counts. The second model based on Ising model uses the energy interaction parameter to characterize a denoised spatial pattern. The Bayesian inference framework allows us to borrow strength in parameter estimation in a de novo fashion. As a result, the two proposed models show competitive performances in accuracy and robustness over existing methods in both simulation studies and two real data applications.
|
