Research Areas

Research

“The formulation of a problem is often more essential than its solution, which may be merely a matter of mathematical or experimental skill. To raise new questions, new possibilities, to regard old problems from a new angle requires creative imagination and marks real advances in science.”

— Albert Einstein & Léopold Infeld, The Evolution of Physics (1938).

Causal Genomics

I develop the conceptual and methodological framework of causal genomics, integrating causal inference with statistical genetics to identify and quantify causal structure in complex genomic systems.

This includes:

• Genome-wide causal interaction

• Multi-omic mediation analysis

• Causal target identification for intervention

• Mendelian randomization for multi-omic data

Causal Inference

My work focuses on identification and inference in complex settings, including:

• Semiparametric theory

• Causal mediation analysis

• Latent/hidden outcomes

• Inference under weak identification

These contributions address fundamental challenges in moving from association to causation.

Mendelian Randomization and Multi-omics Integration

I develop robust Mendelian randomization methods for high-dimensional and heterogeneous genomic data, including:

• Pleiotropy and weak (invalid) instruments

• Inference of complex biological systems

Recent work integrates MR with protein 3D structure prediction (e.g., AlphaFold).

Non-Standard Inference 

I study non-standard inference problems arising in modern data regimes, including:

• Composite null testing in which the classical Central Limit Theorem (CLT) fails 

• Non-standard asymptotics with weak identification

Theory for Deep Learning and AI for Science

I also develop the statistical foundations of deep learning and its integration with scientific discovery in genomics and biomedicine.

• Optimal training time of deep neural networks

• Deep learning for semiparametric estimation and inference 

• Deep learning for causal inference

• AI for proteomics data