Methods exist to appropriately perform association analyses in pedigrees. However, for genome-wide association analysis, these methods are computationally impractical. It is therefore important to determine alternate methods that can be efficiently used genome-wide. Here, we introduce a new algorithm that considers all relationships simultaneously in arbitrary-structured pedigrees and assigns weights to pedigree members that can be used in subsequent analyses to address relatedness. We compare this new method with an existing weighting algorithm, a naïve analysis (relatedness is ignored), and an empirical method that appropriately accounts for all relationships (the gold standard).
Framingham Heart Study Genetic Analysis Workshop 16 Problem 2 data were used with a dichotomous phenotype based on high-density lipoprotein cholesterol level (1,611 cases and 4,043 controls). New and existing algorithms for calculating weights were used. Cochran-Armitage trend tests were performed for 17,333 single-nucleotide polymorphisms on chromosome 8 using both weighting systems and the naïve approach; a subset of 500 single-nucleotide polymorphisms were tested empirically. Correlations of p-values from each method were determined.
Results from the two weighting methods were strongly correlated (r = 0.96). Our new weighting method performed better than the existing weighting method (r = 0.89 vs. r = 0.83), which is due to a more moderate down-weighting. The naive analysis obtained the best correlation with the empirical gold standard results (r = 0.99).
Our results suggest that weighting methods do not accurately represent tests that account for familial relationships in genetic association analyses and are inferior to the naïve method as an efficient initial genome-wide screening tool.