How does PBAT handle extended pedigrees?

When extended pedigrees are used in PBAT, the founders are used in the conditional mean model for the screening process. If siblings and other extended families members are genotyped and phenotyped, then they are used in the calculation on the FBAT statistic (i.e. not the screening procedure). If there is no phenotypic information, but there is genotypic information, then these individuals are used in the conditional mean model to infer founder genotypes when the founder genotypes are not available.

In general there are a limited number of nonfounders that there can be in a pedigree. This number is specified in the "test statistic and computational tab" under "maximal number of nonfounders in one pedigree." Currently, the default is listed under "maximal number of nonfounders in one pedigree" as 14. If there are more than 14 nonfounders in the pedigree, then this number should be increased. As long as this number is greater than the number of nonfounders in all of the pedigrees that are inputted, all pedigrees will stay intact. If this number isn't increased and there are more than 14 nonfounders in the pedigree, then PBAT will automatically split the pedigree up into smaller pedigrees. So regardless of anything, the user never needs to split the pedigrees up.

Please note that it is always preferable to keep large pedigrees in tact rather than splitting them up; however, often extended pedigrees increase the computational time, so it can be impractical to keep all extended pedigrees in tact.

See also: How PBAT splits extended pedigrees can give disparate analysis results.