Validity of downstream analyses for summary statistics from userGWAS

[Emily Leventhal]

Hi all, 

I'd like to know whether it's valid to use factor-derived GWAS summary statistics from GenomicSEM in downstream analyses, and what any considerations may be.

For example, I'm thinking about using the Internalizing Factor summary statistics in:

• Bivariate MiXeR (polygenic overlap between the internalizing factor and another trait)
• LAVA (local genetic correlations), which requires LDSC for sample overlap estimation 
• Mendelian randomization
• Further GenomicSEM models (e.g., mediation model with internalizing factor as the outcome) 

If it is valid, should SNPs with significant Q_SNP be removed before downstream use, or should the full summary statistics be used?

Any guidance, relevant papers, or examples from people who've done this would be very helpful. Thank you for your help!