Between the currently available genotyping panels, aiming for very high coverage at the expense of sample size will result in a decrease rather than an increase of power 19. Choosing a panel with fewer markers will reduce genomic coverage, but the funds saved by choosing the cheaper panel can instead be spent on genotyping more individuals. This will increase the statistical power to a greater extent than adding more markers would do. In addition, recently published methods enable imputation (prediction) of ungenotyped SNPs based on those that are genotyped, utilising LD structures calculated from HapMap 20. This offers a potentially attractive approach for increasing coverage whilst spending more funds on increasing sample size 19. However, imputation is not necessarily a successful solution to increase coverage in all situations. Accurate prediction of an ungenotyped marker depends on the availability of an already genotyped marker to act as a proxy. If a SNP does not have any markers in high LD with it, it cannot accurately be predicted. This means in practice that well-covered regions can easily be filled in but poorly
