Negative beta coefficient (genetic risk score weights) - reversing allele coding

Question

[Juliana]

Hi Dear all,

I've been creating a weighted genetic risk score model, using beta coefficient values [ln(OR)] to evaluate the relative effects (weights) of my interested genetic variants.

However I have the information that when the sign of the beta coefficient is negative, the allele coding is reversed.

Please somebody could help me? What does it mean?

E.g. A SNP coding A/T, the risk allele A. If the beta value for this SNP negative, should I reverse the coding allele for T/A being now the risk allele T, not A? Something related to the DNA strand direction?

I saw that a negative beta value not mean that there is no significant association. And in my data I can see that this is true, because I have SNPs with both negative beta and significant p values. Is it right?

Thank you

Best regards,
Juliana