ABSTRACT
We analyzed some simulated data to assess the success of statistical methodologies to establish the role of the environmental factors (EF) and to identify associated and linked markers. We considered five replicates for each of the four studies, and, with the knowledge of the generating model, concentrated our analyses on chromosomes (CH) 1, 3, and 5. To determine the influence of EF and associated markers on the affection status (AS), we utilized chi-square tests for independence and recursive partitioning (via the CART software). To identify linked markers, we scanned the relevant chromosomes with nonparametric multipoint linkage (NPL) and transmission/disequilibrium tests. These analyses were performed on the whole data set as well as on subsets of individuals and families defined by exposure to EF. CART correctly selected the associated marker (D1G024) and EF1 for Study (ST) 1 and did not generate trees for the other studies. NPL identified the relevant regions on CH3 and CH5 but failed to do so for CH1, except in ST4. Stratifying families by exposure to EF1 did not consistently increase sensitivity of NPL to the relevant CH3 markers, but did help characterize the genetic heterogeneity and identify linked families.
