Are there any advantages of choosing lmer over glmer for my analysis? I am looking to see the association between metabolites and disease. It doesn't necessarily matter whether I have the metabolites as a predictor or as the response, I am just looking to see whether there is a significant association and the direction ( negative or positive effect size). 
I have both fixed effects ( age, BMI, medication usage) and random effects (family ID).
Please could anybody tell me whether there are any benefits over utilizing the lmer over glmer? My disease status is a binary variable ( control as 0, disease as 1). I have run them both and the magnitude of the effect size is different, although for both it is a negative value and the association for both is significant. 
lmer(Metabolites ~ Disease status + Fixed effects ( 7 in total) + random effects 
glmer(Disease status ~ Metabolites + Fixed effects (7 in total) + random effects + family="binomial"
 A: I think the relative merits of the models depends on the research question motivating this analysis, i.e. what do you want to estimate.


*

*the lmer model is best suited to investigate whether the level of metabolites differs between people with and without the disease, and to what extent (after adjusting for your other factors). The estimated difference in the level of metabolites would correspond to the coefficient for Disease status.

*the glmer model would allow you to quantify how the probability that someone has the disease changes with the measured level of metabolites. This seems useful if, say, you are interested in predicting whether one has the disease (making a diagnostic decision?) based on the level of metabolites. Given that you are using a logistic model,the coefficient for Metabolites would describe how the log-odds of having the disease changes with a unitary increase in the level of metabolites.


Hope this helps! I know next to nothing about metabolites and diseases, so I can't say which of these two makes more sense.
