Assistant Professor
Department of Biostatistics
Columbia University
722 West 168th Street
New York, NY, 10032, USA
Email: zl2509@cumc.columbia.edu
Columbia Data Science Institute
Affiliate Member of New York Genome Center
Assistant Professor
Department of Biostatistics
Columbia University
722 West 168th Street
New York, NY, 10032, USA
Email: zl2509@cumc.columbia.edu
Columbia Data Science Institute
Affiliate Member of New York Genome Center
Zhonghua Liu develops the statistical foundations of causal genomics, a term he coined to formalize intervention-level inference in complex genomic systems. His research advances methods for identifying, quantifying, and validating causal structure across high-dimensional genetic and multi-omic data.
Dr. Liu’s work integrates rigorous identification theory with large-scale genomic analysis, spanning Mendelian randomization, high-dimensional causal mediation, interaction-driven genetic architecture, and cross-genome inference. His recent contributions include genome-wide frameworks for host–pathogen causal interaction, structure-aware causal modeling that leverages AI-based protein structure prediction (e.g., AlphaFold), and likelihood-based statistical formulations of neural network training grounded in statistical theory.
By unifying statistical genetics, causal inference, and structured machine learning, his research seeks to move biomedical science beyond association toward mechanistic and intervention-oriented discovery.
His work has appeared in Nature, Nature Computational Science, Cancer Cell, Cell Genomics, Journal of the American Statistical Association, Journal of the Royal Statistical Society Series B, Biometrika, Biometrics, and Annals of Applied Statistics. His research has received international media coverage and has been featured in major institutional and scientific news outlets. He is Principal Investigator of an NIH R01 award and serves as Associate Editor for leading statistical journals.
At Columbia, he leads the Causal Genomics Lab, which develops theory and scalable tools for intervention-level inference across genetic, molecular, and regulatory systems.