Reader Comments

Post a new comment on this article

Mathematical error?

Posted by dahinds on 20 Jan 2012 at 17:52 GMT

The legend for Fig. S2 says, "−log10(4/100,000) = 3.4"

In fact, -log10(40/100,000) = 3.4, indicating that the expected error rate is 10X higher than the value used in the text.

No competing interests declared.

RE: Mathematical error?

sebastiani replied to dahinds on 21 Jan 2012 at 17:00 GMT


thank you for pointing this out. Your observation is correct and the error rate of 4 in 100,000 independent tests relates to a MBF>1,000 and not 100. Sorry for the confusion.

Note however that the selection of SNPs to be used for prediction did not rely on a specific threshold on the expected error rate but rather we used their joint predictive value to inform the model building step.

Paola Sebastiani

No competing interests declared.