iSC MEB

iSC MEB performs integrated batch-effect estimation and HMRF-based spatial clustering on low-dimensional representations of multi-sample spatial transcriptomics (SRT) datasets to delineate spatially structured cellular domains.

Key Features:

• Batch Effect Estimation and Correction: Simultaneously estimates and corrects batch effects across multiple SRT samples during analysis.
• Hidden Markov Random Field (HMRF) Spatial Clustering: Employs a hidden Markov random field model to perform spatially aware clustering of spots or regions.
• Empirical Bayes Framework: Uses empirical Bayes estimation to model parameters and incorporate prior information for improved inference.
• Multi-Sample Integration: Performs joint analysis of multiple SRT datasets to enable comparative and integrative spatial analyses.
• Low-Dimensional Representations: Operates on low-dimensional embeddings of gene expression to reduce dimensionality and computational complexity.

Scientific Applications:

• Mapping Cellular Heterogeneity: Identifies spatial domains and heterogeneity within tissue sections from SRT data.
• Identifying Novel Cell Types or States: Detects spatially localized gene expression patterns indicative of distinct cell types or states.
• Investigating Tissue Architecture in Health and Disease: Compares spatial organization across samples to study alterations in tissue architecture associated with disease or condition.

Methodology:

Empirical Bayes estimation integrated with a hidden Markov random field (HMRF) model, including probabilistic modeling of spatial dependencies and simultaneous batch-effect estimation applied to low-dimensional representations of SRT data.