Statistical methods for 3-dimensional spatial transcriptomics: design, imputation, and analysis

The National Institutes of Health National Human Genome Research Institute awarded The University of Texas MD Anderson Cancer Center $2,759,668 on July 1, 2026, under the Human Genome Research program (CFDA 93.172) to develop statistical methods and computational tools for three-dimensional spatial transcriptomics analysis.

The project addresses analytical barriers that currently hinder biological investigations using 3D spatial transcriptomics by developing three complementary research directions: a robust statistical method to identify the optimal number of two-dimensional slides and locations for experimental design using bivariate P-spline-based methods; a statistical inference method to impute unmeasured three-dimensional spots with quantification of estimation uncertainty; and a three-dimensional-aware dimension-reduction algorithm to detect tissue structure and cellular trajectories within the three-dimensional landscape. These methods will improve experimental design efficiency, sample size, data quality, and visualization capabilities for downstream analysis.

The award is structured as a Project Grant assistance type with a period of performance extending from July 1, 2026, through May 31, 2030. Work is performed in Houston, Texas. The project responds to the urgent need for analytical infrastructure that matches advances in three-dimensional spatial profiling technology, with applications to improving diagnosis and treatment for diseases including cancer and neurodegenerative disorders.

Abstract

Moving from 2-dimensional (2D) to 3D spatial profiling represents a transformative milestone in biological research, offering unprecedented opportunities to enhance the understanding of tissue structure and pathogenesis. IT also holds the promise to improve the diagnosis and treatment for various diseases, such as cancer and neurodegenerative diseases. However, the development of analytical infrastructures and computational tools has fallen seriously behind the technology advances, which present substantial analysis barriers that hinder biological investigations using 3D spatial transcriptomics (ST). The primary objective of this proposal is to address some of the most pressing analytical challenges facing designing and analyzing 3D ST data, including lack of robust strategy to guide the experimental design, lack of tools to recover the ST profiles for cells that are not covered by profiling technology within the 3D structure, and lack of methods to identify structures, microenvironments as well as cellular trajectory. The project is structured around three complementary research directions to overcome the hurdles. 1) we propose a robust statistical method to help identify the optimal number of 2D slides and locations to guide the experimental design for 3D ST profiling. This model works with preliminary or existing 2D ST data, providing cost-effective and statistically optimal designs based on bivariate p-spline based methods. 2) we aim to develop a statistical inference method that can accurately impute the unmeasured 3D spots together with a quantification of estimation uncertainty. This approach can effectively improve the sample size and data quality for downstream visualization and analysis. 3) we will create a 3D-aware dimension-reduction algorithm to detect tissue structure and cellular trajectories within the 3D landscape. The proposed methods will be applied to 3D ST data from gastric cancer patients at the university of texas MD anderson cancer center, as well as lymph node samples from published studies. Although initially motivated by cancer and immunology research, these statistical methods are broadly applicable for 3D ST profiling design and data analysis in other disease contexts and developmental biology studies. Once validated, all software packages developed in this project will be made available to the wider research community. Successful completion of these aims will dramatically enhance the quality and analytical power of 3D ST studies, facilitating the application of these cutting-edge technologies to translational and clinical research.