both snp_r2.exposure and snp_r2.outcome are greater than 1

[ddzhang3]

I came across a strange issue recently, that the correlation between instrument and phenotype became great that the upper limit of 1, when I perform > directionality_test
I dug into the source code, and noticed the following ones:
r_exp <- sqrt(sum(r_exp[!is.na(r_exp) | is.na(r_out)]^2))
r_out <- sqrt(sum(r_out[!is.na(r_exp) | is.na(r_out)]^2))

As tens of SNP were used as instruments, I suspect that these non-weak IVs aggregated as a great-that-one correlation, which lead the NA result of the Steiger test.

Do you have any suggestions?
Thanks