Covariance Matrix for Interaction metrics?

[Luke Cassar]

Hi there, I'm currently trying to carry out studies on gene-environmental interaction. I'm currently trying to use epidemiological stratification and plink's logistic regression with interaction terms separately to compare both methods. My problem is that I want to extract the covariance matrix for each variant so that I can calculate the confidence intervals of RERI (relative excess risk due to interaction) so that I can measure interactions on an additive scale for this method. For an interaction model, it's usually enough to extract the point estimates and transform them. But since I require the confidence interval for RERI, I need the covariance estimates as well, but this output isn't present by default in plink2's --glm output to my knowledge.

Is there a way to extract this somehow without having to make my own logistic/firth regression model in R and running it on each variant?

I also attached a paper here in case the question isn't fully clear regarding what I'm trying to do with transforming the model's coefficients so I can calculate RERI from the interaction model.

Thanks for your time and patience. Any guidance would be greatly appreciated on this.

[Chris Chang]

In the short term, you'd need to run this in something like R, sorry.

In the longer term, an option could be added to --glm to generate an extra output file containing these covariance matrices.  (It would be a poor fit for the usual --glm output file, which can have more than one line per regression.)  How would you want such a file to be formatted?

--
You received this message because you are subscribed to the Google Groups "plink2-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plink2-users...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/plink2-users/aa6c8501-b3a2-40cd-9eef-cb1402d7e0a8n%40googlegroups.com.