Detecting mtDNA effects with an Extended Pedigree Model: An Analysis of Statistical Power and Estimation Bias

Abstract

Mitochondrial DNA (mtDNA) plays a crucial role in numerous cellular processes, yet its impact on human behavior remains underexplored. The current paper proposes a novel covariance structure model with seven parameters to specifically isolate and quantify mtDNA effects on human behavior. This approach uses extended pedigrees to obtain estimates of mtDNA variance while controlling for other genetic and environmental influences. Our Monte-Carlo simulations indicate that a sample size of approximately 5,000 individuals is sufficient to detect medium mtDNA effects (mt2 = 5%), while a more substantial cohort of around 30,000 is required for small effects (mt2 = 1%). We show that deeper pedigrees increase power to detect the mtDNA effect while wider pedigrees decrease power, given the equal total sample size. We evaluated how missing kinship records and mtDNA mutations impact bias. Both lead to underestimation of mtDNA variance, and an overestimation of the interaction between nuclear DNA and mtDNA. In addition, the false positive rate of mtDNA effect estimation is low when fitting the model with data generated without mtDNA effects. Collectively, we demonstrate that using extended pedigrees to quantify the influence of mtDNA on human behavior is robust and powerful.