SMI 2022 - Causal inference in imaging genetic studies

Friday, May 27, 2022 • 3:30–4:45 pm (CT) Light Hall, Room 214

Organizer: Chao Huang, Florida State University

Chair: Jordan Dworkin, Columbia University


Generalized liquid association analysis for multimodal neuroimaging data integration

Lexin Li, University of California, Berkeley

Jing Zeng, Florida State University

Xin Zhang, Florida State University

Multimodal data are now prevailing in scientific research. One of the central questions in multimodal integrative analysis is to understand how two data modalities associate and interact with each other given another modality or demographic variables. The problem can be formulated as studying the associations among three sets of random variables, a question that has received relatively less attention in the literature. In this talk, we propose a novel generalized liquid association analysis method, which offers a new and unique angle to this important class of problems of studying three-way associations. We extend the notion of liquid association of Li (2002) from the univariate setting to the sparse, multivariate, and  high-dimensional setting. We establish a population dimension reduction model, transform the problem to sparse Tucker decomposition of a three-way tensor, and develop a higher-order orthogonal iteration algorithm for parameter estimation. We derive the non-asymptotic error bound and asymptotic consistency of the proposed estimator, while allowing the variable dimensions to be larger than and diverge with the sample size. We demonstrate the efficacy of the method through both simulations and a multimodal neuroimaging application for Alzheimer's disease research.


Imaging genetic-based mediation analysis for human cognition

Rongjie Liu, Florida State University

Tingan Chen, Moffitt Cancer Center

Abhishek Mandal, Florida State University

Hongtu Zhu, University of North Carolina at Chapel Hill

The brain connectome maps the structural and functional connectivity that forms an important neurobiological basis for the analysis of human cognitive traits while the genetic predisposition and our cognition ability are frequently found in close association. The issue of how genetic architecture and brain connectome causally affect the human behaviors remains unknown. To seek for the potential causal relationship, we carried out the causal pathway analysis from single nucleotide polymorphism (SNP) data to four common human cognitive traits, mediated by the brain connectome. Specifically, we selected 942 SNPs which are significantly associated with the brain connectome, and then estimated the direct and indirect effect on the human traits for each SNP. We found out that many of the selected SNPs have significant direct effects on human traits and discussed the trait-related brain regions and their implications.


Mediation analysis with a survival outcome and brain connectivity mediator

Yize Zhao, Yale University

Xinyuan Tian, Yale University

Fan Li, Yale University

It is of great interest to uncover the complex AD pathological mechanism among genetics, neuroimaging, and disease symptoms. Particularly, with a growing body of research on investigate brain function and structure at connectome level rather than voxel or ROI units, understanding how brain connectomes play a role along the genetic to disease pathway becomes crucial. In this work, we develop a general Bayesian mediation framework with networks entailing the mediation effect for a time-to-event outcome. Different from existing mediation analytic approaches which focus on univariate or multivariate mediators, we consider network data as a unified component and impose decompositions to reveal separate sub-network structures along the effect pathways. By impose sparsity, we are able to identify network elements impacted by the exposure and those passing effects to the outcome. We show the superiority of our model in estimating different effect components and selecting active mediating network structures through extensive simulations. By implementing our approach to ADNI2/GO data, we characterize the relationship between the key genetic biomarkers and AD onset mediated by brain structural sub-networks

Return to the schedule page