Table 1.

Comparison of numbers of genes identified as showing strain and/or regional variation in gene expression for different methods

		Number of genes selected at each p-value
Condition	Method	p < 0.001	p < 10-5	p < 10-6
Strain	Sandberg	24	24	24
	ANOVA	213	65	36
	Template	150	56	34
Strain x region	Sandberg	49	49	49
	ANOVA	17	1	0
All strain effects	Sandberg	73	73	73
	ANOVA	230	66	36
	Template	150	56	34
Amygdala	Sandberg	1(3)	1(3)	1(3)
	Template	74	2	0
Hippocampus	Sandberg	2(10)	2(10)	2(10)
	Template	63	16	11
Cortex*	Sandberg	0	0	0
	Template	35	0	0
Forebrain*	Sandberg	58	58	58
	Template	771	191	112
Midbrain	Sandberg	13	13	13
	Template	181	44	23
Cerebellum	Sandberg	150	150	150
	Template	890	339	227
Entorhinal cortex	Sandberg	2(10)	2(10)	2(10)
	Template	37	3	1
All tissue effects†	Sandberg	242	242	242
	ANOVA	874	289	168
	Template	2051	595	374

The number of genes identified by each method, and by Sandberg et al. [1], for each condition or set of conditions is listed if appropriate. For ANOVA and template match, values for three different p-value thresholds are shown. For the various brain regions, the template-match counts include both matches (enriched genes) and anti-matches (genes expressed at lower levels in the region listed). None of the values given is corrected for genes dominated by negative average difference values. After filtering such genes (see Materials and methods) values would be about 5-10% lower. The values in the 'Sandberg' rows are the same in each column because Sandberg et al. did not generate confidence measures for the genes they selected. The totals for 'Sandberg' include genes listed both in Sandberg et al. [1] and Sandberg [12], and group together genes that were 'restricted', 'enriched', 'absent' and 'decreased'. Numbers in parentheses are the totals obtained if one includes genes mentioned in the text of [1] but not listed by name. *Genes specific to the cortex alone. Because Sandberg et al. do not appear to have specifically sought genes expressed only in the cortex (instead they compared cortex only to cerebellum and midbrain [12]), the number of genes in this category for Sandberg et al. are listed as zero. 'Forebrain' is the combination of cortex, hippocampus, entorhinal cortex and amygdala, which most closely matches the 'Cortex' group chosen in [1,12]. It is included here to allow a fair comparison to their methods. ^†This category does not include strain x region effects.