Bias in Gene-Set Analysis Applied to High-throughput Methylation Data

Gene set analysis as it is typically applied to genome-wide methylation assays is severely biased as a result of differences in the numbers and sizes of CpG islands associated with different classes of genes. We demonstrate this bias using published data from a study of differential methylation in lung cancer and a data set we generated to study methylation changes in patients with long-standing ulcerative colitis and show that several of the gene sets that appear enriched would also be identified with randomized data. We also report a method to correct the bias. Application of the corrected method to the lung cancer and ulcerative colitis data sets provides novel and potentially interesting biological insights into the role of methylation in cancer development and chronic inflammation. We used Agilent Human CpG Island microarrays to compare methylation patterns in sigmoid colon tissue between five individuals suffering from ulcerative colitis and five healthy age-matched controls.