Structured Inference, Graphs & Networks for Applied Learning

We develop principled statistical methods for structured and multimodal data with emphasis on graph-based modeling, spatial structures, and high-dimensional analysis in biological applications.

Explore our projects

News & Announcements

We’re recruiting

Our preprints on Conditional(ish) Conformal Prediction and Data Augmentation are now on arxiv!

Sep 2025

Well done to Yating, Yeo Jin, Sowon and Zixuan!

Yeo Jin is giving a talk at the Royal Statistics Society International Conference!

Sep 2025

She will be presenting her work with Yating, Sowon and Zixuan on conformal prediction.

SIGNAL Lab growing

Ongoing

We welcome applications UChicago students (PhD/MS/Undergrad).

What we do

See all projects

GNN Theory & Methods

bias–variance · generalization · model selection

We study statistical properties of GNNs to try and design interpretable, reliable graph learning pipelines.

Structured Estimation

sparse CCA · graph‑constrained models

We develop high‑dimensional estimators that leverage sparsity and known structure for robust inference.

Spatial Transcriptomics

RNA · spatial programs

We deploy our statistical frameworks to uncover spatially organized gene expression patterns and cell–cell interactions from high-resolution transcriptomic data.

Microbial GWAS

thermotolerance · photosynthesis

We use high-dimensional statistical methods to identify genetic and metabolic determinants of key microbial traits, such as thermotolerance and photosynthetic efficiency.