Please read the original paper to understand how results were derived.
A1/A2 is derived from the output of the "PLINK --freq" command, so A1 is the effect allele.
Meaning of columns in the results
beta: refers to regression coefficients where the dependent variable (methylation) was residuals after rank-normalisation to remove outliers and regression on all covariates plus bisulphite-converted DNA (BCD) plate batch to remove potential batch effects (with missing values set to probe mean).
t-stat: t-statistic (the default output of matrix eQTL - beta/standard error)
p-value: this is p-value with no correction for multiple testing Refer to the paper for number of tests and a conservative multiple testing threshold that would be equivalent to p=0.05 if all CpG sites were independent
Effect size: this was requested during peer review as a more biologically interpretable alternative to our regression coefficients (which are on rank-normalised data). Refer to the paper for more detail, but briefly this is the effect sizes as difference in median proportion methylated between homozygote groups.