1
$\begingroup$

I want to investigate, how much sample size i needed to obtain 80% power for GWAS/EWAS studies. Phynotype trait is discrete (not case/control) for human disease.

I wonder, does anyone has came across this? Really appreciate any suggestion or any tool/software which can answer this?

$\endgroup$
8
  • $\begingroup$ Might be worth having a look at this website: csg.sph.umich.edu/abecasis/gas_power_calculator $\endgroup$
    – user438383
    Aug 5 '20 at 14:18
  • $\begingroup$ thanks, but this if for case/control only. Our study is NOT case/control. Any other suggestion? $\endgroup$ Aug 5 '20 at 14:22
  • $\begingroup$ OK - could you edit your question to include some more details about the study you are planning (i.e. quantitative/discrete traits, species, intended outcome etc) and it might make it easier for people to give a good answer to your question. cheers. $\endgroup$
    – user438383
    Aug 5 '20 at 14:24
  • $\begingroup$ I just Edited the question. $\endgroup$ Aug 5 '20 at 14:34
  • $\begingroup$ Do you mean continuous? If it's discrete and there are only two conditions, then the power calculations should be the same as for a case/control study. $\endgroup$
    – gringer
    Aug 5 '20 at 21:45
1
$\begingroup$

Interesting question I would recommend the effective population size $N_{e}$, and error bars around this. Ne can be calculated via $F_{st}$ and is perhaps the most important parameter to determine the genetic population size.

Under the Fisher definition $N_{e}$ is the number of breeding individuals in the population.

The central issue with the approach you are suggesting is that population size does not equate to genetic population size. Thus the population could be huge, but if it is genetically homogeneous it will appear very small. In contrast the number of individuals could be very small, but the genetic variation between between them very large so $N_{e}$ would be huge.

$\endgroup$
2
  • $\begingroup$ Appreciated promt response. Actually, our population is humogeneous, and have very low sample size. I wonder, is there any (genetics related) software where i can play around Ne and Fst? $\endgroup$ Aug 5 '20 at 12:35
  • $\begingroup$ Usual suspects Arlequin etc.. because you have low diversity the traditional stuff will work $\endgroup$
    – M__
    Aug 5 '20 at 12:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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