Fragment yield simulations. (A). The same methylation ratio (1:2) can give different measured ratios for BACs of different GC content. Horizontal lines represent the genomic region of a BAC in a single hybridizing sample. Sample 2 is twice as methylated as sample 1. The upper two rows represent a BAC of higher GC content than the lower rows. Vertical ticks represent unmethylated (digestable) Hpa II sites; check marks denote Hpa II fragments passing the size filter. Hpa II sites methylated in sample 2 but not sample 1 are marked with letters (see text). (B) Simulations based on BAC RP11-47A4. The amount of hybridizing material decreases with increasing methylation. (C) Simulation results for all array BACs. Contour plots summarize the amount of hybridizing material for all BACs at 40% methylation (black contours) and 80% (gray). (D) Predicted ratios when samples of 40% (test) and 80% (reference) methylation are co-hybridized. (E) A more complex scenario results in an opposite relationship between GC content and ratio. In this scenario, each BAC is simulated to report cross-hybridizing material in both test and reference channels equal to 5% of the number of bases of repetitive DNA it contains. (F) Real data for a selected array (male liver:spleen) show that, as predicted, the amount of hybridizing material is higher for higher GC content BACs (liver intensities, black contours; spleen, gray). (G) Measured ratios from the same array (liver:spleen) also have a strong relationship with GC content, as predicted if samples have different overall methylation levels. (H) GC-content-normalized ratios for the same array.