A statistical framework for quantitative trait mapping.

摘要: We describe a general statistical framework for the genetic analysis of quantitative trait data in inbred line crosses. Our main result is based on observation that, by conditioning unobserved QTL genotypes, problem can be split into two statistically independent and manageable parts. The first part involves only relationship between phenotype. second location genome. developed simple Monte Carlo algorithm to implement Bayesian analysis. This simulates multiple versions complete genotype information genomewide grid locations using marker data. Weights are assigned simulated genotypes capture phenotype weighted used approximate quantities needed inference effect sizes. One advantage this approach that weights recomputed as analyst considers different candidate models. device allows focus modeling model comparisons. proposed accommodate interacting QTL, nonnormal multivariate phenotypes, covariates, missing data, genotyping errors any type cross. A software tool implementing procedure available. demonstrate our from mouse backcross population segregating associated with salt-induced hypertension.