R help - Feb 2012 - Logistic regression with non-gaussian random effects

Hi,
does anyone know of an implementation of generalized linear models with random
effects, where the random effects are non-gaussian?

Actually, what I need is to do a logistic regression (or binomial regression)
where the linear predictor in addition to fixed effects and gaussian random
effects also has a term

b*z

where z is a random effect variable with p(z=1)=p(z=-1)=0.5 and b is a parameter
of interest.

I am very grateful to you for any help or comments.

Sincerely,
Line