Hunting genetic diseases: exploring a multistage approach to identifying disease loci

Genomic screen data for a hypothetical disease was used in a two-stage analysis to search for disease loci. We performed both trait-model-dependent and trait-model-free analyses to test their relative power. Results of our first-stage screen in 200 families suggested 13 regions for further analysis. Second-stage follow-up in another 200 families confirmed a single region on chromosome 3 near marker D3G045 with a combined lod score across all 400 families of 6.24 and a sib-pair maximum lod score (MLS) of 4.79. The MLS were highly correlated with both the autosomal dominant and autosomal recessive lod scores in all data sets, suggesting that both trait-model-dependent and trait-model-free methods can be useful for identifying candidate regions for complex disease loci. Reanalysis of the data using alternative sampling schemes suggested that sampling variation has a significant effect on locus detection.