Limitations of stratifying sib-pair data in common disease linkage studies: an example using chromosome 10p14-10q11 in type 1 diabetes.

IDDM10 on chromosome 10p11-q11 has been identified as a putative diabetes susceptibility locus through affected sib-pair (ASP) linkage analysis in UK nuclear families [Davies et al., 1994: Nature 371:130-136; Reed et al., 1997: Hum Mol Genet 6:1011-1016; Mein et al., 1998: Nat Genet 19:297-300]. We extended analysis of linkage to type 1 diabetes in this region by typing a total of 61 markers in a maximum of 418 UK sib-pairs (UK418; peak MLS = 3.84). We then stratified the dataset based on analyses performed previously by both our group [Mein et al., 1998: Nat Genet 19:297-300] and others [Paterson et al., 1999: Hum Hered 49:197-204; Paterson and Petronis, 1999a: Am J Med Genet 84:15-19; Paterson and Petronis, 2000a: J Med Genet 37:186-191; Paterson and Petronis, b: Eur J Hum Genet 8:145-148] and used a permutation procedure to assess the significance of the results. We conclude that the results obtained had a high probability of occurring by chance alone. These data highlight the limitations of stratifying small datasets (n < 500) by additional criteria and the recurrent problems of multiple testing in genetic analysis.