We genotyped 5 genetic variants of different genes and though the risk allele frequencies of all the variants except a single variant did not significantly differ between cases and controls, nonetheless 1 of these variants was found to be associated with the phenotype in the overall study sample including males and females with respect to the over-dominant model. However, 1 variant was found to be associated with the phenotype with respect to the recessive and additive model (but recessive model appeared as the best inheritance model) also in the overall sample. On the other hand, 2 variants were associated with the phenotype only in females but not in males assuming the dominant model. Finally, 1 variant was not found to be associated with the phenotype. Thus, all of the associated variants were showing their effect according to a specific genetic model. However, assuming additive model, only 1 of the 5 variants had a significant p-value (this is the same variant which was also associated assuming the recessive model). But when we made unweighted GRS (sum of the no. risk alleles only), the GRS was observed to be significantly associated with an outcome in an overall population (males and females) showing a very significant per allele effect as well as when groups of GRS were made, the GRS with maximum no. of risk alleles had significance.

Now, I have one confusion that all the associated variants showed association with respect to different model but they showed significant association with phenotype when combined in the GRS, however, the GRS is made on the basis of no. of risk alleles (a kind of additive model) which are not significantly different in cases vs. controls. Moreover, the individual variants were either associated in the overall study sample or only in females but GRS was associated with the whole sample population.

Can you please help me on how to explain and resolve this problem? I really tried to find a reason but could not find any appropriate explanation and reference for the issue.

Looking forward to the responses.

Thank you in advance.