Quality control steps involved removing cases and SNPs with a high degree of missingness, and removing SNPs with a minor allele frequency less than 5%. We also performed identity by state (IBS) analysis to identify unknown relatives or duplicates and multi-dimensional scaling (MDS) to identify those with differing ethnic backgrounds to the majority of the group (a Caucasian cohort). The WTCCC2 controls clustered tightly together with our cases in the MDS plot, showing that our cases and controls were ethnically compatible (Appendix 1). A total of 387 cases (from 400 genotyped) and 5,380 controls remained for analysis in the discovery sample after these steps.
