I am working on the GWAS analysis of some quantitative traits that are not normally distributed. I applied a log transform on some of the traits, but some of the traits still do not appear normal. Out of all the methods of normalization (Box-Cox, log, rank transform, and quantile normalization), are there pros and cons in choosing particular methods of normalization? Do they greatly affect the results of the analysis if I quantile-transformed a dataset to a normal distribution with standard error of 1?