Overview: FAMILY Procedure

Family genotype data, though more difficult to collect, often provide a more effective way of testing markers for association with disease status than case-control data. Case-control data can uncover significant associations between markers and a disease that could be caused by factors other than linkage, such as population structure. Analyzing family data by using the FAMILY procedure ensures that any significant associations found between a marker and disease status are due to linkage between the marker and disease locus. This is accomplished by using the transmission/disequilibrium test (TDT) and several variations of it that can accommodate different types of family data. One type of family consists of parents, at least one heterozygous, and an affected child who have all been genotyped. This family structure is suitable for the original TDT. Families having at least one affected and one unaffected sibling from a sibship that have both been genotyped can be analyzed using the sibling tests: the sib TDT (S-TDT) or the nonparametric sibling disequilibrium test (SDT). Both types of families can be jointly analyzed using the combined versions of the S-TDT and SDT and the reconstruction-combined TDT (RC-TDT). The RC-TDT can additionally accommodate families with no unaffected children and missing parental genotypes in certain situations.

When the trait of interest is quantitative, regression and variance component analyses can be used to test for marker associations (Allison 1997; Fulker et al. 1999; Rabinowitz 1997). These models were extended to accommodate any size nuclear family with or without parental genotypes (Abecasis, Cardon, and Cookson 2000; Monks and Kaplan 2000) and then to general pedigrees (Abecasis, Cookson, and Cardon 2000). The strength of many procedures in SAS/STAT in these areas can be applied to these statistical tests, though some data manipulation is required to form the correct inputs. In order to simplify the data preparation steps, PROC FAMILY can produce an output data set containing the pair of allelic transmission scores at each marker allele. This data set can be used in the MIXED procedure, for example, to test for association and linkage between marker genotypes and a quantitative trait via the method of Abecasis, Cookson, and Cardon (2000).