Family-based association tests for qualitative and quantitative traits using single-nucleotide polymorphism and microsatellite data.

Using the Genetic Analysis Workshop 12 simulated data, we contrasted results for association tests in nuclear families and extended pedigrees using single-nucleotide polymorphism (SNP) data, and we compared results for different trait definitions, for outbred and isolate populations, and for SNP and microsatellite data. SNPs in major genes 1 and 6 were analyzed using transmission disequilibrium testing (TDT) [Spielman et al., Am J Hum Genet 52:506-16, 1993], sibship disequilibrium testing (SDT) [Horvath and Laird, Am J Hum Genet 63:1886-97, 1998], family-based association testing (FBAT) [Horvath et al., Eur J Hum Genet 9:301-6, 2001], and a chi-square analysis of founders. TDT and SDT were applied in a sample of independent nuclear families, while FBAT was applied in extended pedigrees. SNPs and microsatellites were analyzed with dichotomous and quantitative trait definitions using FBAT in the isolate and outbred populations. The results of the TDT, SDT, and FBAT analyses are comparable using SNP data to identify the disease gene. However, these tests of association were not helpful in discriminating between functional and non-functional SNPs in disequilibrium. SNP data were able to identify association with affection status in a gene that influences the liability directly (MG6), but did not perform as well when assessing association with affection status in a gene that influences the outcome only through a quantitative trait (MG1). Association with MG1 was observed using the SNP data when the outcome was defined quantitatively. Microsatellite data were relatively unsuccessful in identifying association with the markers in the region of a major gene. The magnitude of the associations between SNPs and the dichotomous or quantitative trait definitions were similar in the outbred and isolated populations.

Influence of marker heterozygosity and genetic heterogeneity on fine mapping.

The purpose of the current study was to utilize the Genetic Analysis Workshop 12 simulated data to evaluate fine-mapping strategies for quantitative traits. We approached the analysis as if it was a follow-up to a genome scan that had identified two regions of interest and used the provided 1-cM density microsatellite typing data to mimic fine mapping of these regions. As these investigators knew the true locations of the putative genes under study, we explored the effects of the informativeness of microsatellite markers (marker heterozygosity) and the effects of genetic heterogeneity across families for ten replicates of the data. These results shed a cautionary light on the reliability of fine-mapping efforts on refining mapping locations as the position and the strength of the lod score can be markedly affected by the sampling of the population, the amount of variation accounted for by the gene, and the informativeness of the marker. Our studies did not reveal a large effect of unlinked families on the shape of the lod score peak.