step2_pos.R - problem with the Z score of TWAS

[MinghanQ]

Hi,
I am currently attempting to apply the SUMMIT method for TWAS analysis.
While studying the software source code, I noticed that in 'step2_pos.R’, on line 271, the p-value is calculated by dividing z.twas by the square root of r2.twas. According to the formula in your original article, it seems that this value is the Z score for TWAS. However, on line 272, z.twas is directly output as the Z score. Why is this the case? Could it be a misunderstanding on my part?"

[melothemightyone]

Hi MinghanQ,

Your observation is keen and correct. The object z.twas / sqrt(r2.twas) is the Z-score that should be outputted. We will fix this issue in an updated version of SUMMIT in the near future.

[lpgilchrist]

Hi,

I am also interested in applying SUMMIT when I came across this GitHub issue. Am I to understand that on line 272 where currently the code for defining the TWAS z-score is update[10 + w] <- z.twas it should be:

update[10 + w] <- z.twas / sqrt(r2.twas)

This issue is also present in the step2_rsid.R code on line 251.

Thanks in advance for your help

[melothemightyone]

Hi,

I am also interested in applying SUMMIT when I came across this GitHub issue. Am I to understand that on line 272 where currently the code for defining the TWAS z-score is update[10 + w] <- z.twas it should be:

update[10 + w] <- z.twas / sqrt(r2.twas)

This issue is also present in the step2_rsid.R code on line 251.

Thanks in advance for your help

Hi lpgilchrist,

Your observation is correct. Please revise the two codes accordingly.

[lpgilchrist]

Great, thanks for clarifying! I will update the code accordingly.