Unconditional two-locus nonparametric linkage analysis

Lars Ängquist, Dragi Anevski and Holger Luthman

Centre for Mathematical Sciences
Mathematical Statistics
Lund Institute of Technology,
Lund University,

ISSN 1403-9338
We discuss different aspects of unconditional two-locus nonparametric linkage (NPL) analysis with special emphasis on gene-gene interaction. We interpret this as identical-by-descent (IBD) sharing correlation between two disease loci both having marginal effect. We relate this to the concept of two-locus NPL score functions, the possible importance of using a composite rather than a simple null hypothesis and the corresponding calculation of statistical power. Moreover, we define several classes of score functions and give multiple suggestions on how to incorporate a composite null hypothesis into the analysis. The least favourable two-locus IBD-distribution is discussed, resulting in an upper bound of the two-locus p-value.
Key words:
NPL analysis, unconditional two-locus linkage analysis, genetic disease models, IBD-sharing, gene-gene interaction, score functions, composite null hypothesis, least favourable distribution, Monte Carlo simulation, estimation of genetic parameters, power calculations