Hello Everyone,
I have a dataset of ~400 individuals typed at 14 microsatellites (sampling was individual-based), and would like to analyze these using RDA. I initially estimated Smouse and Peakall (1999) genetic distance among individuals, but I’m not sure I could use a matrix of genetic distances as a response variable for RDA analyses.
Alternatively, I thought of using the frequency/count of the most common allele (per locus) in genotypes instead. Ex. with one locus and 2 alleles:
Alleles = ‘A’ and ‘B’, ‘A’ being the most common among the individuals sampled.
Individual—genotype -- ‘A’ allele count in genotype:
Ind1--AA--2
Ind2—AB-- 1
Ind3—BB— 0
Could this be a better/adequate way of representing the response variable?
Thanks in advance for your help.
Dear Jimena,
You can construct a PCA-based genetic matrix of genotypes as explained in Shirk et al.
(2010). See paper attached for an example of use of RDA in this context.
All the best
Julien
Dear Jimena,
You can use a genetic distance matrix as a response variable in a RDA. This corresponds to a dbRDA (see Legendre and Anderson 1999). You can perform a Principal Coordinates Analysis (PCoA) of your distance matrix and use the principal coordinates as the response variables.
Thomas
http://onlinelibrary.wiley.com/doi/10.1890/0012-9615(1999)069%5B0001:DBRATM%5D2.0.CO;2/abstract
Hi! Took me a while to figure this out but maybe it could be useful to someone else or someone may prove me wrong!
1. So I first got my genepop file and transformed it into a genind using the function "genepop_to_genind()" in R.
2. After that I calculated a matrix of euclidean distances on the genind using the function "mat_gen_dist()" #you can specify the type of distance you prefer, I went for "basic" which is Euclidean.
3. I obtained a matrix of euclidean distances between my populations! This dissimilary matrix can only be used as input of the function capscale() (so distance-based RDA) in the package vegan, as the function rda() allows genotypic data encoded as 0,1,2.
4. Performed the capscale() with an environment with variables (nrow= n of pop, n.col=variables).
I hope this helps!! (AND I HOPE THIS IS RIGHT!)
Giulia
Hi! Took me a while to figure this out but maybe it could be useful to someone else or someone may prove me wrong!
1. So I first got my genepop file and transformed it into a genind using the function "genepop_to_genind()" in R.
2. After that I calculated a matrix of euclidean distances on the genind using the function "mat_gen_dist()" #you can specify the type of distance you prefer, I went for "basic" which is Euclidean.
3. I obtained a matrix of euclidean distances between my populations! This dissimilary matrix can only be used as input of the function capscale() (so distance-based RDA) in the package vegan, as the function rda() allows genotypic data encoded as 0,1,2.
4. Performed the capscale() with an environment with variables (nrow= n of pop, n.col=variables).
I hope this helps!! (AND I HOPE THIS IS RIGHT!)
Giulia
- Badges
- Science topic