The Relative Power of Family-Based and Case-Control Designs for Linkage Disequilibrium Studies of Complex Human Diseases I. DNA Pooling

Abstract

We consider statistics for analyzing a variety of family-based and nonfamily-based designs for detecting linkage disequilibrium of a marker with a disease susceptibility locus. These designs include sibships with parents, sibships without parents, and use of unrelated controls. We also provide formulas for and evaluate the relative power of different study designs using these statistics. In this first paper in the series, we derive statistical tests based on data derived from DNA pooling experiments and describe their characteristics. Although designs based on affected and unaffected sibs without parents are usually robust to population stratification, they suffer a loss of power compared with designs using parents or unrelateds as controls. Although increasing the number of unaffected sibs improves power, the increase is generally not substantial. Designs including sibships with multiple affected sibs are typically the most powerful, with any of these control groups, when the disease allele frequency is low. When the allele frequency is high, however, designs with unaffected sibs as controls do not retain this advantage. In designs with parents, having an affected parent has little impact on the power, except for rare dominant alleles, where the power is increased compared with families with no affected parents. Finally, we also demonstrate that for sibships with parents, only the parents require individual genotyping to derive the TDT statistic, whereas all the offspring can be pooled. This can potentially lead to considerable savings in genotyping, especially for multiplex sibships. The formulas and tables we derive should provide some guidance to investigators designing nuclear family-based linkage disequilibrium studies for complex diseases.

Reporting, Appraising, and Integrating Data on Genotype Prevalence and Gene-Disease Associations

“…A possible solution is pooled analysis (see below). One research strategy proposed for the future is large-scale testing by genome-wide association mapping (52,60,64,65). It is important to note that this strategy is hypothesis generating rather than hypothesis testing, and thus it may require additional safeguards against type 1 error.…”

Worldwide mutation spectrum in cartilage-hair hypoplasia: ancient founder origin of the major70A→G mutation of the untranslated RMRP

“…Two million Monte Carlo replicates using different seeds were computed for the 2.1 Mb correspondence, and the average likelihoods taken to represent the final values because of the poorer convergence to a definite maximum as compared to the 1 Mb correspondence to genetic length of 1 cM, where only one million replicates were used. We also computed the means-of-moments estimator for the most recent ancestor (TMRCA) 22 to evaluate the agreement with DMLE age estimates. We did not apply the correction as suggested by Labuda et al 23 Markers TESK1 A/G SSCP, delG, D4S, AC1, and XL1S were selected for the estimation, since they had more than one instance of presumably nonancestral allele presented in the haplotype in disease chromosomes.…”

Lack of replication of thirteen single-nucleotide polymorphisms implicated in Parkinson's disease: a large-scale international study

“…Although this design has the advantage of being most robust to population stratification, it can be underpowered relative to case-unrelated control studies. 17 With 332 matched case-unrelated control pairs, tier two was also underpowered to detect ORs in the very low range. Limited power may also result in falsenegative findings.…”