I am performing GWAS on simulated data following this tutorial Hail | GWAS Tutorial.
In particular:
hl.import_plink(bed = InputFile + '.bed', bim = InputFile + '.bim', fam = InputFile + '.fam', quant_pheno = True).write(InputFile + '.mt', overwrite=True)
mt = hl.read_matrix_table(InputFile + '.mt')
gwas_res = hl.linear_regression_rows(y=mt.quant_pheno,x=mt.GT.n_alt_alleles(),covariates=[1.0])
My understanding is that the linear_regression_rows function perform a linear model of the type:
Y = XB + e where y is the non-standardised quantitative phenotype, X is the individual genotype and e is some error.
I am interested to perform GWAS without using a genetic relatedness matrix, which I think is achieved by using linear_regression_rows instead of linear_mixed_model (for instance). Correct?
Also, in my command line above I used covariates=[1.0], should I use covariates=[0.0] instead? Or this does not make any difference except for the intercept of the linear regression (which will be equal to the average phenotype )?
Thank you in advance,
Gabriele