Ziyi Li
University of Texas MD Anderson Cancer Center, TX·Mathematics & Statistics
New lab technologies can now measure which genes are switched on at each location within a slice of tissue. This method is called spatial transcriptomics (ST).
- Fields
Mathematics & Statistics
Computer Science
Medicine & Clinical Research
- Topics
spatial transcriptomics
experimental design
imputation
- Grants
- 3 known grants, through Jul 2030
What they work on
- Develop a statistical method based on bivariate P-splines that uses existing 2D spatial transcriptomics data to choose the optimal number and placement of tissue sections, making 3D experiments cost-effective.
- Create statistical methods that impute gene activity at unmeasured 3D tissue locations with uncertainty estimates, plus a 3D-aware dimension-reduction algorithm to find tissue structures and cell trajectories in gastric cancer and lymph node data.
- Build a statistical model that estimates how cell-cell interactions vary across a tissue section, accounting for the mix of cell types in each spot and its location, and compare these patterns across patients.
- Extend cell-cell interaction models to samples collected over time and to 3D data that combine serial sections with tissue images, applied to pancreatic and gastric cancer and Alzheimer's disease brain samples.
- Develop transfer-learning Cox survival models, including subgroup-specific versions, that borrow information from external registries missing key variables to improve risk prediction in inflammatory breast cancer and myelodysplastic syndromes.
More about the research
It shows how cells are organized and how they influence one another in diseases such as cancer and Alzheimer's disease. The statistical tools needed to design these experiments and make sense of the data, however, have not kept up. Clinical data present a related problem: rare cancers often have too few patients at a single hospital for reliable risk prediction. The research here develops statistical methods and free software for both problems, using data from patients at MD Anderson and from published studies.
Why it matters
Better tools for 3D spatial transcriptomics could let cancer and neuroscience researchers design cheaper experiments and find the cell interactions linked to disease and patient outcomes. These findings could point to early-detection markers and new therapeutic targets. For patients with rare, aggressive cancers such as inflammatory breast cancer and myelodysplastic syndromes, more accurate risk prediction could help clinicians choose treatments suited to each patient.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Statistical methods to delineate the spatial and temporal pattern of cell-cell interactions in Spatial Transcriptomics data
NIH · Sep 2025 – Jul 2030
Statistical Methods for 3-Dimensional Spatial Transcriptomics: Design, Imputation, and Analysis
NIH · Jul 2026 – May 2030
Addressing Unmeasured Covariates in Source Cohorts with Transfer Learning for Survival Outcomes
NIH · Jun 2026 – May 2028
Closest labs
Yi Li
University of Michigan, MI·Mathematics & Statistics
Develop integrative causal machine learning methods that pool multiple high-dimensional observational cancer cohorts, using the Boston Lung Cancer Survival Cohort and the International Lung Cancer Consortium, to compare treatments despite unbalanced patient characteristics across sites.
Akihiko Nishimura
Johns Hopkins University, MD·Mathematics & Statistics
Develop statistical methods that combine Bayesian hierarchical models, biomedical ontologies and high-dimensional transfer learning to pool genetic and clinical data from sites that differ in patients, data coding and practice.
Hongzhe Lee
University of Pennsylvania, PA·Mathematics & Statistics
Develop transfer learning methods that borrow information across ancestry groups and tissues to improve polygenic risk score prediction and increase power to detect trait-associated variants in minority populations.
The school
University of Texas MD Anderson Cancer Center: 23 labs on PI Pages, $1.4B a year on research.
Reach out
A start for your email. Fill in the brackets.