Depends on multiple factors, primarily the size of the population and size of the genome. As your genome size increases, the more markers you need to adequately cover the genome because markers will no longer be linked with the phenotype. Similarly as your population size increases, the amount of recombination events being observed will increase and so linkage between markers with break down and more markers will be needed. If you have a small population and a small genome, millions of SNPs would be overkill since many markers would be in linkage disequilibrium, look identical, provide no additional information and no extra resolution.

The shanun. I am working on wheat. I have 150 Italian genotypes that were genotyped using 500 DArT markers and would like to run GWAS. I did not find any study about the minimum number. I run GWAS and found the markers that were previously published with the same trait.

Dear Gregor,

With the bigger genome size (in humans, animals, or plants), you need more and more positions to detect significantly associated variants. It depends on your pre-specified distance through the genome for selecting the SNPs. But the thing is, you or the machine that generated the data should evenly pick up the positions through the genome. This condition, evenly distributed, is more important than the number of SNPs.

Besides, If the number of SNPs in your research is low, you will expect fewer variants that are significant signals or even new! In other words, the chances of winning will decrease.

Hopefully helpful.

Best,

try this paper. you may find the answer.

Article Linkage disequilibrium and within-breed genetic diversity in...