Is it possible to apply SEM to data derived from exonic sources?

[mike]

Dear all,

I would like to inquire about the application of SEM (Structural Equation Modeling) in processing exon associations. Specifically, my data consists of GWAS results for common variants with a frequency greater than 0.005, and exon association analysis for rare variants and gene burden.

I am considering filtering out variants with a frequency less than 0.005 and retaining those sourced from exome_array_rollup and imputed_topmed. Although exome_array_rollup lacks INFO, I can retain variants based on a MAF greater than 0.005. Is this approach appropriate?

Thank you for your assistance.

Best regards

[mike]

The specific article can be viewed at https://www.nature.com/articles/s41586-021-04103-z#Abs1. It conducts separate regression analyses for exome data and imputed data, but combines them for downstream analysis. Consequently, some GWAS lead SNPs originate from exome data. For Structural Equation Modeling (SEM), should I exclude the exome-derived data from the summary statistics file and retain only the imputed data? I am inclined to keep the exome-derived data (with MAF > 0.005) and combine it with the imputed data for SEM analysis. I would appreciate your advice.