|
Activity Number:
|
158
- SPEED: Statistical Methods, Computing, and Applications Part 2
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 8, 2022 : 10:30 AM to 11:15 AM
|
|
Sponsor:
|
International Society for Bayesian Analysis (ISBA)
|
|
Abstract #323754
|
|
|
Title:
|
Bayesian Modeling of Spatial Molecular Profiling Data at the Single-Cell Level
|
|
Author(s):
|
Jie Yang* and Sunyoung Shin and Qiwei Li
|
|
Companies:
|
The University of Texas at Dallas and University of Texas at Dallas and The University of Texas at Dallas
|
|
Keywords:
|
Negative binomial mixture model;
Spatial transcriptomics;
Ising model;
Energy function;
Double Metropolis-Hastings algorithm
|
|
Abstract:
|
Recent technology breakthroughs in spatial molecular profiling (SMP) have enabled the comprehensive molecular characterization of cells while preserving their spatial and gene expression contexts. One of the fundamental questions in analyzing SMP data is to identify spatial variable (SV) genes whose expressions display spatially correlated patterns. Existing approaches are built upon either the Gaussian process-based model, which relies on ad hoc kernels or the energy-based Ising model, which requires the gene expression measured on a lattice grid, where each unit captures up to hundreds of cells. We develop a generalized energy-based framework to model the gene expressions measured for each cell, which is irregularly distributed. Our Bayesian model includes a negative binomial mixture model to dichotomize the raw count data to reduce the noise and a geostatistical marking model with a generalized energy function, in which the interaction parameter is used to identify the spatial pattern. We use auxiliary variable MCMC algorithms to sample from the posterior distribution with an intractable normalizing constant. We demonstrate our method on both simulated and real data.
|
Authors who are presenting talks have a * after their name.
