Improved ChIP-chip analysis by a mixture model approach

BMC Bioinformatics

Table 3 Comparison of different methods for the simulated data set with 5,100/4,998 spike-in regions for unamplified/amplified samples, respectively

	Mixer (Lowess)		Mixer (MA2C)		MA2C		TileMap (p = 0.01, q = 0.05)		TileMap (p = 0.08, q = 0.13)		HGMM
Sample	D	FDR	D	FDR	D	FDR	D	FDR	D	FDR	D	FDR
GSM254930	4359	0.16	5753	0.28	4829	0.19	2775	0.001	3872	0.003	3707	0.02
GSM254971	4969	0.23	5110	0.23	4697	0.22	2758	0.001	3682	0.005	3615	0.02
GSM254972	5135	0.22	3957	0.19	4738	0.19	2978	0.001	4114	0.011	3558	0.03
GSM254973	4714	0.18	4795	0.20	4560	0.20	2695	0.001	3534	0.003	3493	0.02
GSM254805	4537	0.25	4784	0.27	3860	0.22	1946	0.003	2744	0.022	2237	0.03
GSM254806	4878	0.21	5826	0.32	4284	0.17	2672	0.0004	3924	0.022	3085	0.03
GSM254807	4957	0.21	5157	0.24	4569	0.20	2487	0.0004	3802	0.003	3508	0.02

See main text for the simulation methods. About 10.2% of the probes are from spike-in regions.

ISSN: 1471-2105