西交

The Grid template

Shiquan Sun, PhD

Professor, Department of Biostatistics, Xi'an Jiaotong University; Director, Center for Single-Cell Omics and Health

The Grid template Shaanxi, China

The Grid template Twitter

The Grid template Github

The Grid template Google Scholar

The Grid template Orcid


Research



Cell Atlases

How can we build reference maps that work across studies and still distinguish rare cell populations and tissue-specific states? Single-cell and spatial multi-omics datasets differ in what they measure, at what resolution, and in which tissues. We develop methods that align comparable cell populations across datasets, resolve finer cell states within each tissue, and quantify uncertainty in cell annotation and spatial mapping. These atlases allow us to distinguish molecular changes within a cell population from shifts in the abundance and spatial organization of cell populations in disease.


Multiscale Modeling

How can we estimate what would happen if an intervention were started earlier or a different treatment were used? Molecular profiles typically capture snapshots, clinical measurements are irregular, and treatment decisions often depend on a patient’s evolving health. We develop causal models of cells, organs, and individuals that use time-resolved perturbation data to constrain how biological states evolve and longitudinal clinical data to estimate how those changes relate to functional decline. We account for time-varying confounding and quantify uncertainty when comparing intervention strategies.


Applications

How can we detect early functional decline, and why do people with the same diagnosis follow different trajectories? We use cell and tissue atlases to identify molecular signatures, then test whether those signatures precede measurable decline and improve prediction beyond routine clinical measurements. We develop models that update individual risk as new measurements become available and use independent cohorts to test whether molecular subtypes predict distinct trajectories. Where intervention data are available, we assess whether those subtypes also differ in treatment response.