News for hackers. Stuff that matters.
Re:5-HTTLPR: A Pointed Review by outlace on Thursday May 09, 2019 @01:33PM (#19870284)
I wonder if this has anything to do with the problem with multiple comparisons. In a GWAS since you're comparing possibly thousands of genes against controls, the probability of getting a false positive is higher just by the multiple tests you're running.

So you have to correct for this often by dividing the chosen p-value threshold (usually 0.05) by the number of comparisons (this is the simple Bonferonni correction). If you looked at 10,000 genes your new significance threshold would be 0.05/10000=0.000005. Therefore any gene difference between groups must have a p-value of less than 0.000005 to be considered significant if you're testing that many genes. However, if you chose to only look at 5-HTTLPR, then you don't need this correction and you only need p < 0.05.

There are no comments on this comment yet.