As a precursor to the aforementioned imputation-based strategies, Mukherjee et al. (2011) investigated combining control samples for only those SNPs genotyped on all arrays to be combined. They showed that combining publically available data sets based on only the SNPs shared across arrays is a non-biased technique that can substantially improve statistical power as the control:case ratio increases as long as ancestry stratification and MAF variation across control data sets are properly accounted for (Mukherjee et al. 2011). However, this approach substantially limits coverage of the genome which is likely to reduce statistical power through limited opportunities to test variants at or associated with causal loci (Spencer et al.2009).
