|
Abstract:
|
Whole-genome association scans involve hundreds of thousands of genetic markers, selected from more than 4 million known variants. The scale of the problem is unprecedented in genetic research. A number of methods have been developed for marker selection, most of which are aimed at exploiting known patterns of correlations in reference samples to minimize redundancy. These design choices have a number of potential consequences on subsequent analysis strategies. Missing data in experimental conditions, which can occur at high rates and generally not at random, exacerbate the problem. We describe methods for marker selection and examine the trade-off between efficiency and restrictions in subsequent analysis and computational challenges facing this novel area of applied genetic research.
|
