Obtaining Accurate P-Values from SNP-Based Linkage Scans

• Activity Number: 277
• Type: Topic Contributed
• Date/Time: Tuesday, July 31, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Ryan Subaran*+ and William C.L. Stewart
• Companies: Columbia University and Columbia University
• Keywords: Linkage Analysis ; Genetics ; Type I error

Abstract

When linkage methods are used to detect disease genes, the evidence against the null hypothesis must exceed some accepted threshold of significance (e.g. LOD > 3.0). However, since this approach rarely accounts for linkage disequilibrium (LD), the type I error is often inflated. In fact, when data were simulated under an empirical pattern of LD, this approach inflated the type I error by more than 12%.

Therefore, to facilitate the accurate estimation of p-values for any SNP-based linkage study and for any test statistic, we created a new simulation program that uses a Markov model along with multilocus inheritance to generate genotypes for all family members under the null hypothesis of no linkage, but with any desired level of LD. As such, our program Haplodrop, accurately estimates the p-value, and controls the type I error. Furthermore, in the few situations where existing software can be directly compared to Haplodrop, the results agree. Haplodrop is freely available from the web as part of the EAGLET software package for genetic analysis.