Robust estimation of critical values for genome scans to detect linkage

Authors

  • Silviu-Alin Bacanu

    Corresponding author
    1. Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    • Department of Psychiatry, 424 Thomas Detre Hall, 3811 O'Hara Street, Pittsburgh, PA 15213
    Search for more papers by this author

Abstract

Estimation of study specific critical values for linkage scans (suggestive and significant thresholds) is important to identify promising regions. In this report, I propose a fast and concrete recipe for finding study specific critical values. Previously, critical values were derived theoretically or empirically. Theoretically-derived values are often conservative due to their assumption of fully informative transmissions. Empirically-derived critical values are computer and skill intensive and may not even be computationally feasible for large pedigrees. In this report, I propose a method to estimate critical values for multipoint linkage analysis using standard, widely used statistical software. The proposed method estimates study-specific critical values by using Autoregressive (AR) models to estimate the correlation between standard normal statistics at adjacent map points and then use this correlation to estimate study-specific critical values. The AR-based method is evaluated using different family structures and density of markers, under both the null hypothesis of no linkage and the alternative hypothesis of linkage between marker and disease locus. Simulations results show the AR-based method accurately predicts critical values for a wide range of study designs. © 2004 Wiley-Liss, Inc.

Ancillary