Calibration for inverse-variance-weighted meta-analysis

[shinichinamba]

Hi, thank you for developing this tool.

From my understanding, the calibration appears to be applied only to the p-values, while the effect sizes and standard errors remain unchanged.

For inverse-variance-weighted meta-analysis, what would you recommend?

• Should we use the original (uncalibrated) beta and SE values?
• Or should we apply the calibration to either the beta or the SE before performing the meta-analysis?

Thank you.

[hrushikeshloya]

Hi,

You are right we only calibrate the p-values and the test statistic.

For your purpose, you can apply the calibration to beta and calculate the SE using the test statistic and beta. This is similar to what FastGWA does to estimate the effect sizes and SEs. See Section 3 (page 4) of the FastGWA Supplements.

Hope this helps.