domains / families / science / genomic-sem

genomic-sem

separate

Fits structural equation models to GWAS summary statistics using GenomicSEM (Grotzinger 2019), including common-factor models, confirmatory factor models, ESEM, common-factor GWAS with Q_SNP heterogeneity, multivariate Wald tests, and stratified GenomicSEM partitioned heritability. Reconciles results against MTAG multi-trait analysis. Handles sample overlap via the LDSC sampling-covariance matrix, identifies and resolves Heywood cases, and verifies model fit with CFI / RMSEA. Use when modeling latent genetic architecture across correlated traits, running multivariate GWAS on a shared factor, distinguishing factor-mediated from trait-specific SNP effects, or comparing GenomicSEM common-factor results against MTAG when both depend on accurate sampling covariance.

When to use

• Fitting structural equation models to GWAS summary statistics
• Modeling latent genetic architecture across correlated traits
• Running multivariate GWAS on a shared factor
• Distinguishing factor-mediated from trait-specific SNP effects
• Comparing GenomicSEM common-factor results against MTAG
• Identifying and resolving Heywood cases
• Verifying model fit with CFI / RMSEA
• Stratified GenomicSEM partitioned heritability

When not to use

• Tasks not involving GWAS summary statistics
• Single-trait analysis without latent factor modeling

Skill page

Source