GWAS & Logistic regression in R

Question

[carlos.maldonadomota]

Hi

I have my genotype data and my phenotype data, I made GWAS with GAPIT tool, but now I would like to use the logistic regression analysis in R, which software is the appropriate ? What would be the first step to work with the data?

Thanks.

Answer 1

[Jeremy Leipzig]

A logistic regression is where your dependent, outcome variable is dichotomous or binary (aka logical).

GAPIT is a mixed model linear regression, which means it tries to divide the regressors into fixed and random effects. This is super important in agriculture where most of all the GWAS signals end up being SNPs associated with some inbred line.

They correct for this by this assigning fixed effects to SNPs and call the population a random effect, or vice-versa.

GAPIT assumes a continuous linear outcome variable. There is such a thing as mixed model logistic regression. Actually, I wouldn't be surprised if GAPIT has a logistic method hiding somewhere.

So...is your outcome a number or a category?

[carlos.maldonadomota]

It is a number (0,1)

[Jeremy Leipzig]

Is it continuous?

[carlos.maldonadomota]

No,I just have 0 and 1

[Jeremy Leipzig]

It's probably still ok to use GAPIT for this, you just have to round the responses to 0 or 1, assuming these responses are ordered in some meaningful way