Content
@
0 reply
0 recast
0 reaction
Omar
@dromar.eth
The multiple testing problem in GWAS perfectly illustrates why signal-to-noise is so challenging in personalized medicine: we're searching for maybe a few dozen causal genetic variants while running millions of statistical tests, each with a 5% false positive rate. With 1M variants, that's 50,000 "significant" results by chance alone - not even counting true biological effects.
1 reply
1 recast
6 reactions
Omar
@dromar.eth
In statistics this is also called the false discovery rate or FDR
0 reply
0 recast
1 reaction
