Table 1

Algorithm power
Sample size BioTile TileMap CHARM BioTile vs. TileMap BioTile vs. CHARM
5 0.06 0.56 0.09 OR = 0.05, p = 4.6 × 10-182 OR = 0.60, p = 1.2 × 10-03
10 0.83 0.48 0.27 OR = 5.45, p = 5.1 × 10-81 OR = 14.66, p = 1.1 × 10-196
15 0.89 0.48 0.44 OR = 8.34, p = 1.7 × 10-112 OR = 9.83, p = 5.9 × 10-132
20 0.93 0.67 0.49 OR = 5.17, p = 7.7 × 10-52 OR = 11.70, p = 4.2 × 10-133
DMR length (# probes)
5 0.82 0.25 0.15 OR = 11.39, p = 8.9 × 10-29 OR = 20.14, p = 7.2 × 10-40
10 0.91 0.60 0.24 OR = 3.64, p = 1.1 × 10-07 OR = 19.92, p = 2.6 × 10-39
15 0.93 0.72 0.39 OR = 4.86, p = 1.0 × 10-08 OR = 17.82, p = 2.6 × 10-32
20 0.97 0.78 0.62 OR = 9.25, p = 2.4 × 10-09 OR = 22.21, p = 2.4 × 10-22
25 0.95 0.83 0.69 OR = 3.65, p = 1.2 × 10-04 OR = 7.12, p = 3.5 × 10-11
30 0.99 0.84 0.77 OR = 18.65, p = 2.2 × 10-08 OR = 29.94, p = 2.9 × 10-13
Log2 fold change
0.1 0.74 0.36 0.35 OR = 4.93, p = 7.7 × 10-19 OR = 5.18, p = 7.0 × 10-20
0.5 0.90 0.58 0.36 OR = 7.26, p = 1.1 × 10-14 OR = 18.17, p = 1.8 × 10-33
0.75 0.97 0.57 0.37 OR = 20.17, p = 1.1 × 10-10 OR = 41.03, p = 2.4 × 10-18
1 0.97 0.83 0.50 OR = 7.90, p = 5.0 × 10-08 OR = 43.53, p = 3.5 × 10-41
1.5 1.00 0.85 0.63 OR = 55.96, p = 2.9 × 10-14 OR = 194.39, p = 1.9 × 10-41
2 1.00 0.88 0.75 OR = Inf, p = 1.4 × 10-03 OR = Inf, p = 4.0 × 10-07

A table depicting the power of each algorithm to identify ‘hidden’ DMRs inserted into the simulated data matrix. Fisher’s odds ratios over 1 denote a higher proportion of DMRs identified by BioTile relative to TileMap and CHARM, respectively.

Guintivano et al.

Guintivano et al. BMC Bioinformatics 2013 14:76   doi:10.1186/1471-2105-14-76

Open Data