multiple groups scripts for ACE models

3 replies [Last post]
AnneN.'s picture
Offline
Joined: 02/27/2014

Hi all,
For my research, I ran several univariate and multivariate ACE models using path specification and now I have to extend these models to do multiple groups (young and old) for these very same analyses. I understand the logic of creating a heterogeneity model first and then equating the estimates for both groups in the homogeneity model. I have been trying to modify the non-ACE multiple groups scripts taken from the OpenMx User guide but I have run into many problems. I also tried Hermine's heterogeneity scripts but I am unfamiliar with the matrix approach and so I haven't had much success with this either. Does anyone out there have a sample multiple- groups script with path specification, either for a univariate or a multivariate ACE that I could look at to model after? Thank you in advance for your help.
Sincerely,
Anne

RobK's picture
Offline
Joined: 04/19/2011
Offhand, I don't know of an

Offhand, I don't know of an example script that does what you describe (though there may well be one out there).

Let me make sure I understand what you're looking for. You want to do an twin-ACE analysis, using path specification, and you have two groups, young and old. Correct? If so, then your overall mxModel is going to contain four submodels (groups): young DZ, young MZ, old DZ, old MZ. If the phenotype(s) is(are) something where it makes sense to separately model males and females, you would instead have eight groups: young DZ males, young DZ females, young MZ males, young MZ females, old DZ males, old DZ females, old MZ males, old MZ females.

If you want, you could post the script you're trying to turn into multigroup, or post your "rough draft" attempt at writing it, and I (or someone else) could give you some suggestions for how to proceed.

AnneN.'s picture
Offline
Joined: 02/27/2014
Thanks, Rob, for your offer

Thanks, Rob, for your offer to help. I ended up spending a couple of weeks working on this and I was able to get the univariate model to run. Phew!! I still have to extend it to the multivariate but it looks like I now have a handle on where to create the sub models and the relevant changes that need to come with it. I will keep you posted if I have troubles with the multivariate nested model.
Thanks again,
Anne

RobK's picture
Offline
Joined: 04/19/2011
Good to hear

OK, that's an encouraging sign! Glad you're making progress figuring everything out on your own.