This is something we dreamed up (see refs below) to explore how good a marker would be at detecting a disease association caused by a different marker. This metric imagines that one marker of a pair acts to increase disease susceptibility with a relative risk (RR) of 10, and calculates the RR that we'd expect to find if we typed the second marker.
For a pair of markers A and B:
RR10M1 denotes the RR we expect to find for marker B if marker A has RR=10. It is given by
{f
B x ([10 x f
ab] + f
Ab)} / {f
b x ([10 x f
aB] + f
AB)}
RR10M2 denotes the RR we expect to find for marker A if marker B has RR=10. It is given by
{f
A x ([10 x f
ab] + f
aB)} / {f
a x ([10 x f
Ab] + f
AB)}
References
- Ackerman et al (2003) Haplotypic analysis of the TNF locus by association efficiency and entropy
- Burgner et al (2003) Nucleotide and haplotypic diversity of the NOS2A promoter region and its relationship to cerebral malaria
