A structural equation model for imaging genetics using spatial transcriptomics

Sjoerd M H Huisman,Marcel J T Reinders,Boudewijn P F Lelieveldt,Ahmed Mahfouz,Nematollah K Batmanghelich

doi:10.1186/s40708-018-0091-0

Sjoerd M H Huisman, Marcel J T Reinders + Show 3 more

Open Access

https://doi.org/10.1186/s40708-018-0091-0

Copy DOI

Abstract

Imaging genetics deals with relationships between genetic variation and imaging variables, often in a disease context. The complex relationships between brain volumes and genetic variants have been explored with both dimension reduction methods and model-based approaches. However, these models usually do not make use of the extensive knowledge of the spatio-anatomical patterns of gene activity. We present a method for integrating genetic markers (single nucleotide polymorphisms) and imaging features, which is based on a causal model and, at the same time, uses the power of dimension reduction. We use structural equation models to find latent variables that explain brain volume changes in a disease context, and which are in turn affected by genetic variants. We make use of publicly available spatial transcriptome data from the Allen Human Brain Atlas to specify the model structure, which reduces noise and improves interpretability. The model is tested in a simulation setting and applied on a case study of the Alzheimer’s Disease Neuroimaging Initiative.

Highlights

1 Introduction The aim of imaging genetics studies is to find associations between genetic variants and imaging features, often in a disease context [1]. This scheme extends beyond traditional genome-wide association studies (GWAS) by identifying genetic associations of imaging biomarkers with the assumption that these biomarkers are a more direct reflection of the genetic effects
We propose a method to identify associations between candidate genetic variants and imaging features allowing for the incorporation of prior knowledge
2.1 Variables used We model the relationship between single nucleotide polymorphisms (SNPs) and brain region measurements

Summary

Introduction

The aim of imaging genetics studies is to find associations between genetic variants and imaging features, often in a disease context [1]. One of the largest imaging genetics studies [10] analysed over 30,000 individuals within the Enhancing Neuro Imaging Genetics through Meta-Analysis (ENIGMA) consortium They performed a genomewide association of SNPs with seven brain volumes and identified only eight genome-wide significant SNPs. Despite the high dimensionality of the imaging data (millions of voxels), the actual number of independent tests for which we need to correct in an imaging genetics study is far smaller than the number of voxels. These models consist of a directed acyclic graph, which can be made to incorporate covariates, including possible confounding factors Both of these studies use relatively small candidate SNP sets, because they aim for understanding SNP–brain relationships rather than the discovery of genome-wide associations. For these reasons the region groups in the dimension reduction are based on spatial gene expression data of the brain

Materials and methods

Results

Conclusion