How does PBAT define and use informative families?

When using PBAT, the number of informative families depends on genotypic information and what genetic model you are using. PBAT defines an informative family as "Having at least one parent contributing to the variance of the offspring genotype during transmission." This is based on the assumption that transmission follows Mendelian rules.

This means that for most models, an informative family occurs if you have at least one heterozygous parent. Having two homozygous parents gives you exactly what the offspring genotype will be.  FBAT tests rely on the comparison between expected transmission and observed transmission of alleles.  With two homozygous parents there is nothing to compare and the family is uninformative.

However, with a recessive model this is a little different. For example - a recessive trait in which (1_1) corresponds with having a disease. If you had one homozygous (0_0) parent and one heterozygous (0_1) parent, there is still no way that you can have an affected offspring (1_1). Therefore, even with one heterozygous parent it would not be an informative family.

In your PBAT results you are given the number of informative families for each SNP.  This is the actual number of families that were used to calculate your FBAT test statistic and the corresponding p-value.

There is a setting in which you define a minimum number of informative families (by default 10).  If PBAT scans through your data and does not find 10 informative families for that SNP, it will not do the analysis.  If you ever get a message stating "Execution Finished.  There Was No Spreadsheet Output" it often means that you did not meet the minimum number of informative families.  If you set the number to 0 all analysis will be performed regardless of the number of informative families.