My goal is to work on GWAS method development as a research project. However, human genome data are usually confidential because of the identification problem, so it's very hard to get them. (application takes a long time and I may not get it at the end.)

Therefore, maybe I can try to play with the data of other species first. So I wonder where I can find some GWAS data for Zebrafish and rice. (The reason I am asking for these two is that I find researchers tend to perform genome analysis to mice, Arabidopsis Thaliana, zebrafish, and rice, and I already found where to download the GWAS data for the first two.)

I can find several data sets (like ZFIN) that have the annotated genome information, but that's only one genome. Where can I download the raw data (like genotypes of 500 zebrafishes and corresponding phenotypes) to start my own analysis?

  • 1
    $\begingroup$ Possible duplicate of Where can I find a database that has phenotype information together with associated SNPs? $\endgroup$ Jul 19, 2017 at 9:01
  • $\begingroup$ @KamilSJaron Thanks. In my defense, the previous question is asking for the associated results that people already find with their own research. This question, I'm looking for a data set that I can research on. What are the differences between human and these species? I did this mainly because it take a long time to apply for human genome data and we may not get it. $\endgroup$ Jul 19, 2017 at 12:57
  • 3
    $\begingroup$ You do not need to defend yourself, I am trying to help you in making question clearer. I would recommend you to edit your question to add there more details (including those you wrote in comment). Tell us what you want, what you have tried and why you think it did not work. If the problem with human is dataset size, why not just subsample human dataset? However, GWAS is notoriously known for requirement of huge sample sizes. $\endgroup$ Jul 19, 2017 at 13:20

1 Answer 1


I think it will be more productive for you if you choose one of the public human datasets and practise your GWAS methods on that. A couple of examples examples:

  • 1000 genomes -- use "population" as your trait of association. You'll find that you get association across the whole genome, so the analysis method would change from "what bits are associated" to "what are the bits that are most consistently associated"

  • NCI Genomic Data Commons -- open access (and controlled access) data associated with cancer

  • $\begingroup$ I think you need bit more precise phenotype information than population, if you want to test a potential of method. For population you can just look as 2. and 3. axis of PCR and it is going to work (you do not need any formal statistics)... $\endgroup$ Jul 20, 2017 at 8:49
  • $\begingroup$ But that's not going to tell you what markers are associated. I've had success in the past doing a GWAS-like analysis on populations. $\endgroup$
    – gringer
    Jul 20, 2017 at 20:13
  • $\begingroup$ You can look at the projection of loci to principal components and therefore you see what loci are actually responsible for the separation. I do not doubt that you can associate loci with populations. I am trying to say that you do not need any sophisticated method for that because the signal is very strong. I had an impression that the challenge of GWAS studies lays in identification of weak signals, therefore for benchmarking taking "easy" data is not really useful (maybe for sanity tests). $\endgroup$ Jul 21, 2017 at 9:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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