ABSTRACT
Several meta-analytic techniques have been developed for combining information from multiple studies in contexts other than linkage detection. We apply the technique of combining parameter estimates to the problem of finding disease loci in the simulated data and compare results with those obtained by reanalyzing pooled raw data. To facilitate the combination of study results, we highly recommend that parameter estimates and their standard errors be reported in published studies. If different research groups were to make original data available, progress toward disease gene location and characterization may be more quickly made.
Subject(s)
Genetic Linkage , Models, Genetic , Genetic Testing , Genetic Variation , Genome , Humans , Models, StatisticalABSTRACT
Meta-analysis has been little explored to make an overall assessment of linkage from different studies. In practice, it is likely that published linkage studies will only report p-values. We compared the performance of the widely used Fisher method for combining p-values with that of pooling raw data. More loci were consistently found by pooling raw data. In the absence of further information, combining p-values can provide an overall, but limited, assessment of different linkage studies. However, meta-analysis would be better viewed as a preliminary step toward the goal of analyzing the pooled raw data.