The Cascade model (described here) is a latent variable model that uses extended twin family data to simultaneously estimate several different environmental and genetic contributors to phenotypic variation. It is an offshoot of the Stealth model, which is the brainchild of Lindon Eaves, Andrew Heath, Mike Neale, and K. Truett among others, and has been developed and extended largely by Hermine Maes over the last 15 years. If one has extended twin family data (twins, siblings, twin spouses, twin parents, and twin children), it probably makes sense to use either the Stealth or Cascade model because they tend to produce estimates that are much less biased and more precise than estimates from alternative, simpler models, such as the Classical Twin Design. The advantage of the Cascade over the Stealth is that it can model several different modes of assortative mating and vertical transmission. We are currently working on a paper that quantifies the bias and precision of the Cascade model, and compares these to the bias and precision of the Stealth, Nuclear Twin Family, and Classical Twin designs.
The Cascade algebra can be found here and the Cascade Mx script, written in large part by Sarah Medland, is here. We are currently working on porting this Cascade script to the OpenMx software, which should vastly simplify and shorten the code. A nuclear twin family model (the heart of the Cascade model) for two variables (bivariate) and written in OpenMx is here.
The code for the Stealth script, written by Hermine Maes, is here. There are two advantages to the Stealth script over the Cascade presently. First, it is about twice as fast as the Cascade (taking ~ 10 mins for 70K subjects). Second, it is written to be able to model multivariate data.
The Mx code for a basic Nuclear Twin Family design is here and the code using OpenMx is here.