If you refer to the "beta", it represents what the regression coefficients would be if the model were fitted to standardised data. In other words, it represents the "slope" of the regression line and therefore it can have negative or positive values. In practical terms, if we talk about eQTLs assuming an additive model, a positive beta means that the tested allele is associated with an increase expression of that gene; a negative beta means that the tested allele is associated with a reduction in gene expression.