Preprocessing differential methylation hybridization microarray data

Table 5 Ovarian cancer AUC measurement table

(1): p < 0.05		none	sub	edwards	normexp	normexp50
	none	0.758	0.754	0.754	0.732	0.742
	LOESS	0.802	0.812	0.819	0.820	0.823
	composite	0.764	0.771	0.771	0.782	0.796
	control	0.805	0.826	0.820	0.811	0.803
(2): p < 0.04		none	sub	edwards	normexp	normexp50
	none	0.755	0.750	0.750	0.744	0.748
	LOESS	0.809	0.798	0.807	0.817	0.823
	composite	0.767	0.785	0.779	0.791	0.788
	control	0.808	0.832	0.827	0.810	0.802
(3): p < 0.03		none	sub	edwards	normexp	normexp50
	none	0.761	0.750	0.750	0.765	0.754
	LOESS	0.811	0.822	0.822	0.819	0.816
	composite	0.781	0.792	0.792	0.780	0.772
	control	0.802	0.817	0.816	0.802	0.802
(4): p < 0.02		none	sub	edwards	normexp	normexp50
	none	0.765	0.755	0.758	0.748	0.757
	LOESS	0.840	0.821	0.828	0.823	0.798
	composite	0.783	0.776	0.776	0.767	0.777
	control	0.791	0.807	0.809	0.808	0.811
(5): p < 0.01		none	sub	edwards	normexp	normexp50
	none	0.767	0.772	0.776	0.741	0.751
	LOESS	0.806	0.800	0.800	0.803	0.794
	composite	0.781	0.773	0.773	0.767	0.760
	control	0.802	0.816	0.815	0.794	0.791

Each row is for one p-value cutoff point. Within each row, the first column contains a p-value cutoff point; the second column contains a sub-table of 20 numbers corresponding to the AUC measurement results of 20 preprocessing methods. The underlined bold numbers are the top 3 largest numbers in each sub-table of the second column.

ISSN: 1756-0381