I have been using Sambada but with my quantity of loci (arround 1700) together with the fact that the software applies Bonferoni correction, I am only able to find associations when I drop my loci number to 15....
Do you know other programs where I can look for Genotype - Environmental variable associations?
I'm thinking of trying PLINK and treating the environmental variables as phenotypic variables, what do you think of this approach?
Hi!! So there are some papers published regarding association with NGS data with environmental variables (see: http://www.biomedcentral.com/1471-2164/15/801). In that paper they used both GAPIT and LFMM. GAPIT is a software for classical GWAS and they treated the environmental variables (transformed by PCA) as phenotypes. Instead LFMM (http://membres-timc.imag.fr/Olivier.Francois/lfmm/index.htm) has been specifically designed for gene-environment interaction, and it's included in the R package LEA (http://membres-timc.imag.fr/Olivier.Francois/LEA/index.htm), that is an integrated framework for landscape genomic analysis that includes several conversion tools (from vcf, hapmap and so on).
Hope this help.
PLINK is a good start! If there is also a spatial aspect to your study (say you sampled populations in different geographic areas), you might wish to look at BayEnv (http://www.genetics.org/content/early/2013/07/03/genetics.113.152462 ). It corrects for background population structure divergence due to geographic separation.
Dear Telma
Apart from the wonderful suggestions above, PLATO could be a wonderful alternative for PLINK, especially when you pose problems with TPED input files: http://plato.stanford.edu/entries/eugenics/
Best
Prash
All the previous recommendations are spot-on. Here a recent review that might come handy: A practical guide to environmental association analysis in landscape genomics http://onlinelibrary.wiley.com/doi/10.1111/mec.13322/abstract
Just an update ont his topic:
I ended up using BayEnv2.0 which I then replaced for BayPass which builds up on BayEnv2.0 incorporating more into its models. BayPass is relatively user friendly if you have some background experience with such software and the manual & paper are very complete.
Both can take a matrix with more than 1000 loci. Just check if your environmental variables are properly normalized, otherwise your results may be biased (as you'll be able to see in the resulting graphs) I had to normalize mine with a 0-1 scoring method to have reliable results.
These are currently (from what I've read in various papers and reviews) the best around, since they consider population structure and diminish possible confounding effects of demography: important things to avoid false positives.
Me and coworkers did some scripts to prepare inputs for both BayEnv and BayPass that migth be usefull:
from geste format to BayEnv: https://github.com/Telpidus/omics_tools/blob/master/GESTE2BayEnv.py
BayEnv results miner: https://github.com/CoBiG2/RAD_Tools/blob/master/bayenv2_results_miner.py
from geste format to BayPass:https://github.com/CoBiG2/RAD_Tools/blob/master/geste2baypass.py
BayPass workflow in R: https://github.com/CoBiG2/RAD_Tools/blob/master/Baypass_workflow.R
- Badges
- Science topic
- Similar topics
- Psychology