Greetings! How is the beta-coefficient (effect column) which is used for METAL input computed? Thanks. Alessandro
Hi Alessandro, so, if you are using METAL, I guess you need to perform meta-analysis of association studies. Most software performing association studies will provide as a result not only the p-value, but also a beta-estimator, or an estimate of the effect of the risk allele. This is what you need to input as beta-coefficient.
From the METAL wiki: "The EFFECT column can have positive and negative values (beta values from regression, for example), or simply directions of effect relative to the reference allele, listed as “+” and “-“. An EFFECT of “+” (or any positive number) with respect to the reference allele A (or effect allele A), for example, represents a case where increasing number of copies of allele A are correlated with increasing trait values. For discrete traits, it is common to report odds ratios, which are always positive. In this case, to calculate the direction of effect, one should look at the log of the odds ratio. METAL can compute the odds ratio for you if you specify EFFECT log(ODDS_RATIO_COLUMN)"
Good luck!
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Thanks Fabio. I guess I can estimate the beta effect from the OR by calculating its natural logarithm. I supposed I had to do so, but I wasn't sure because SNPtest output file reports a beta effect which is different from the logarithm of the odds ratio.
Indeed, they are different, because the beta is the slope of a regression while the odds ratio not. However, METAL is willing to take LOG ODDS RATIO and use it as the EFFECT. I don't know METAL very well, but I guess it just wants a directional measure of the effect of the risk allele, be it a LOG ODDS RATIO or a beta coefficient of regression to perform the meta-analysis.
Thanks again, Fabio. Although the directional measure of effect can be computed with ln(OR) and METAL accepts it, I tried to figure out how to compute the beta coefficient regression, just for the sake of accuracy, but I didn't come up with a formula.
The beta coefficient is computed by the software used for genetic association. Not all the software packages do that. Which software did you use? For example Genabel does that computation.
Is it better to use base 10 or base e logarithm to estimate beta from odds ratio?
Hello. If I use log(OR) as EFFECT, do I need to convert OR's standard error (SE) as well?
usually not. the SE should be for log(OR). but please manually compute the P value to double check.