Login

News

ACE, ADE or ACDE model

Posted on
No user picture. valentinav Joined: 06/15/2020
Forums
Behavioral Genetics Models
Hi All,
I saw papers in which they estimated ADE instead of ACE because the correlation between MZs was twice as big as the one for DZs.
So far in my analysis I used the ACE model, and I was wondering whether I should try the ACE one.
I also read of some analysis estimating ACDE model, is there any example twin I could look at?
Is the ACDE analysis impossible when twins are raised together?

Thank you very much for your help

Replied on Wed, 01/27/2021 - 11:32
Picture of user. AdminRobK Joined: 01/24/2014

I also read of some analysis estimating ACDE model, is there any example twin I could look at?
Is the ACDE analysis impossible when twins are raised together?

See my answer here.

Log in or register to post comments
Replied on Thu, 01/28/2021 - 05:49
No user picture. valentinav Joined: 06/15/2020

Thanks!
I had seen the paper you mentioned before but I am struggling with find a script to perform this analysis.
In the paper they mention this link http://www.010.upp.so-net.ne.jp/koken/bg.html. but apparently it does not function anymore.
Is there any other script available estimating the ACDE?

Thanks!

Log in or register to post comments
Replied on Tue, 02/02/2021 - 10:47
Picture of user. AdminNeale Joined: 03/01/2013

The ACDE model identification from MZ/DZ reared together classical twin study relies on skewness/kurtosis of the data to infer the presence of non-additivity. For most behavioral/psychological phenotypes (and many others), such interval level of measurement does not exist. Application of the method to poorly scaled variables seems likely to cause high rates false positive and false negative findings. Far better would be to add a different type of relative than just MZ/DZ twins. Same-aged adoptees reared together or separated twins would be good. Non-twin data usually have the added issue of unequal age/cohort effects.

Note that the C in the ACE model really subsumes a lot of things: regression on age, effects of assortative mating, genuine C, non-additive genetic (dominance & epistasis) factors, and some GxC interaction, among others. See Verhulst & Neale on the merits of estimating C without a lower bound of zero. When one allows C to go negative, any negative C estimates are telling you that the balance of dominance/epistasis to C is at the dominance/epistasis end of the seesaw.

Log in or register to post comments
  Login

News