I would like to do some gene inheritance simulations, and I was wondering if there is an accepted model for what mating looks like in the general human population? I am not necessarily looking for the exact graph of any population.

For example, my current model has generations 1, 2, ..., n, n+1, ... where each individual in generation i is the product of a mating between two individuals in generation i-1. This is obviously over simplistic, and I was wondering if there are better models out there?


Well, there are some questions regarding what population. I don't have 50 rep, so I can't really ask for clarifications; populations range a lot, from your typical avoidance of inbreeding (laws against cousin marriage) to Tamils with their high rates of first cousin marriage (so high that they don't have any pedigree inbreeding depression due to selection, so it's not unhealthy).

Mating in humans is often assortative for many traits including education, vocation, etc. When the Americas were settled (after aboriginal Americans were devastated by disease), skin color was a source of assortative mating. Arguably it still is. On the other hand, people tend to mate disassortatively for traits such as Major Histocompatibility Complex. Furthermore, isolation in human populations tends to be by distance (though this is of course changing, leading to a lot of admixture in recent population history).

As a result, what your population is will have a huge impact on your results; you'll need to calculate f for any particular locus, and take a look at admixture. If f is 0, then you have a random mating scenario (neither preference nor avoidance of inbreeding, and neither assortative nor disassortative mating).

TL;DR if you're doing a very simple analysis on dummy data, you can probably get away with assuming that mating is totally random. But if we're talking about actual populations, you'll need to run the models/calculations of f and other population measures.

  • $\begingroup$ My analysis applies first to the United States. Perhaps I can generalize after that. It is okay if subpopulations are neglected as long as I can make get good results for the vast majority. I might just start with a random mating model as you suggest. $\endgroup$ – Paul Aug 1 '14 at 1:15
  • $\begingroup$ On the one hand, there's papers such as this that discuss assortative mating. On the other hand, you have things like this that indicate some level of disassortative mating as well -- but note that that only applies to some of the populations. If you can't do any analysis of f values, then you could also try doing a statistical test of f; for instance, run a simulation using _f_=0 and compare to your data. $\endgroup$ – Henry Gong Aug 1 '14 at 17:13

An unbounded model with no carrying capacity(unregulated reproduction) will be easier to model, but it would have no long term resemblance to actual population change. It is easier to model by virtue of one not being required to know the carrying capacity. That being said, the carrying capacity is important in a realistic model since the population is naturally regulated by external factors. You might do well to look at Discrete time logistical model. The reason I suggest this model is that it is discrete and recursively defined, thereby lending itself to represent each iteration as a likely branch in the graph.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.