Statistical and AI Models in Spatial Omics for Precision Medicine and Health

Spatial transcriptomics enables high-resolution measurement of gene expression while preserving the spatial layout of a tissue. A key question is how to identify genes that vary across space in ways that align with the underlying tissue structure. Standard approaches often use the raw two-dimensional coordinates of the measurements, but these can miss patterns when the biology is organized along effectively one-dimensional structures, such as layers or gradients. 

Advances in genomic profiling in situ now allow the measurement of molecular activity across intact tissues with exceptional breadth and resolution. These data, obtained from spatial sequencing, high-plex imaging, or omics-enhanced digital pathology, reveal how cellular states are shaped by their local environments. I will present work addressing key challenges in spatial biology: quantifying cell–cell interactions, uncovering spatially coordinated molecular gradients, and achieving power-preserving data reduction that enables analysis at unprecedented scale. Building on these developments, I will show how causal inference can be applied to spatial omics to perform in silico perturbations of tissue neighborhoods, predicting how the molecular landscape would change if a particular cell type were added or removed. Together, these developments bring us closer to predictive tissue biology, where computational models can anticipate how cellular interventions reshape tissue function. 

"Spatial niche" is widely used in spatial omics yet inconsistently defined, obscuring resolution, components, and analytic purpose. We summarize how the term maps to three distinct concepts: (i) a proximity-based neighborhood around an anchor cell/spot; (ii) a computationally derived spatial domain (mutually exclusive or overlapping) grouping spots with shared molecular or contextual features; and (iii) a biologically grounded microenvironment that includes cellular and non-cellular elements (e.g., vasculature, ECM, ligands) supporting resident cells. We organize methods and datasets under this taxonomy and clarify two application modes: using niches as covariates for spot-level modeling versus using niches as spatial clusters for within- and between-niche inference (composition, differential expression, ligand–receptor signaling). We provide reporting standards and selection guidance, and recommend reserving "niche" for the biological microenvironment while using "neighborhood" and "spatial domain" for the other cases to reduce ambiguity and enable rigorous, comparable analyses. 

An essential first step in the analysis of spatial transcriptomics data is to assign cell types to each spatial location. This process is complicated by the presence of cell type mixtures on individual spatial locations. The best performing cell type identification algorithms are based on supervised methods that rely on a reference dataset to estimate cell type expression profiles. However, finding a high quality annotated single-cell RNA-seq (scRNA-seq) reference dataset is difficult and often impossible. Here, we address this challenge by developing an unsupervised factor-based statistical method for identifying cell types in spatial transcriptomics datasets, which we call Reference-free Inference of Cell types and Expression (RICE). We model gene expression as a linear mixture of cell type-specific gene expression profiles, and both cell type proportions and cell type-specific gene expression are estimated via maximum likelihood within our probabilistic model. We demonstrate, in several Slide-seq and MERFISH spatial transcriptomics datasets, RICE's accuracy in estimating both cell type proportions and cell type-specific gene expression. We show that RICE achieves comparable accuracy to state of the art supervised methods when a scRNA-seq reference is available, while it can outperform these methods when the reference is less reliable due to cell type-specific platform effects. We further show that our sparse factor modeling approach outperforms existing non-sparse unsupervised factor-base methods. We distribute RICE within the R package \url{https://github.com/dmcable/spacexr}.