Sign up ×
Biology Stack Exchange is a question and answer site for biology researchers, academics, and students. It's 100% free, no registration required.

I have been doing a lot of reading on quantitative genetics and the G- (and B-) matrix lately. I get the principle behind performing the analysis now but I am still not sure how to do it. I'd like to be able to try it out with some dummy/old data in R to see how it is done.

Let's pretend I have measured 5 traits of Drosophila (Wing length, lifespan, eye pigmentation, bristle number, and fitness). Could I construct a G-matrix for the population from this data and also get a B-matrix to measure the selection on the traits?

Aims of the post:

1) What information (trait measurements, fitness scores etc.) is needed to construct G- and B- matrices?

2) I would like to find print or online material that would guide me in actually implementing this method. It seems like there is an awful lot of papers saying that G-matrices are great, but no one actually says how to do them in real life...

share|improve this question
Your link talk about G-matrix only. I don't know what a B-matrix is. Do you have a link that explain what is a B-matrix? Thks – Remi.b Aug 24 '14 at 19:15
@Remi.b The b-matrix is a sub matrix within the g-matrix (the g matrix is constructed of four sub matrices, Gm [male genetic variance and covariance] Gf for female too, B (and B transposed) with the cross sex genetic covariance – rg255 Aug 24 '14 at 20:55
Ok, that makes sense to me. Thank you! – Remi.b Aug 24 '14 at 20:57
A short description of what is a G-matrix can be found in this question – Remi.b Sep 5 '14 at 23:24

1 Answer 1

up vote 0 down vote accepted

To construct the G-matrix you need additive genetic variances and covariances for all traits, so you normally need results from breeding experiments (e.g. phenotypic midparent-offspring data), that you can do parent-offspring regressions on. Don't know any good online sources but see Balding et al. 2007 p. 534ff for some info. I've seen methods that claim that the G-matrix can be estimated directly from phenotypic data from unrelated individuals (e.g. Zintzaras 2011) or from genomic info e.g. SNPs (Vattikuti et al. 2013), but are not familiar with these and do not know how reliable they are.

A clear background on using mixed models to estimate the G-matrix, including examples/tutorials, can be found in Wilson et al (2009).

Or did you have something else/something more specific in mind?

share|improve this answer
There is really two aspects to this for me I guess. Firstly, I'd like to know how they can be done. Secondly I am trying to figure if I could do them using trait values from hemiclonal lines of drosophila such that I get sex specific trait values for each haplotype and can therefore make a B matrix. – rg255 Oct 18 '13 at 11:42
Well, you should be able to get genetic variances from hemiclonal lines - you need to carefully consider exactly how to calculate them and the amount of heritability captured though. What do you mean with: "Firstly, I'd like to know how they can be done". What the matrices contain, how to technically peform the regressions or something else? I suspected that you were after something more specific, but you should update the question to make it less unclear. – fileunderwater Oct 20 '13 at 22:11
Also, have you looked at Animal models (statistical mixed models), where the entire pedigree can be used to estimate components of the G-matrix? – fileunderwater Oct 20 '13 at 22:23
what I mean by "how they can be done" is what sort of data do I need, what needs to be measured, as it already says in the question – rg255 Oct 21 '13 at 7:08
@GriffinEvo I guess this is what is unclear to me; in the simple case, what you need for a G-matrix is phenotypic data on the traits of interest and a known relationship between individuals (e.g. parent-offspring). I also wrote this in my answer, but apparently you are looking for something else. To account for e.g. environmental factors, sex or a known pedigree you need a more complex statistical model, e.g. Animal models. I have updated my answer with a paper that might be of use. – fileunderwater Oct 21 '13 at 8:57

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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