Table 2.

hcrtr and pharmacological classifier performance Download Figure 4-1, TIF file.

Comparison	Motifs (number)	Cv error (%)	Cv error Std	Mc error (%)	Mc EP	RM error (%)	RM error Std
hcrtr
Day and night
•WT/Het	173	25.5	6.77	27.7	1.88	39.0	1.93
•WT/Hom	83	24.7	6.07	50.0	2.83	47.8	3.66
•Het/Hom	235	24.7	3.76	27.7	1.88	38.6	1.83
Day
•WT/Het	80	19.5	9.60	27.7	2.66	37.9	1.08
•WT/Hom	195	16.7	7.50	50.0	4.00	48.7	4.22
•Het/Hom	55	22.7	7.02	27.7	2.66	33.8	2.01
Night
•WT/Het	79	16.3	6.38	27.7	2.66	37.1	7.17
•WT/Hom	53	12.8	9.58	50.0	4.00	52.3	6.16
•Het/Hom	76	16.0	7.27	27.7	2.66	36.0	5.05
Melatonin (day)
•Control	40	0	0	25.0	4.42	16.4	3.67
•0.01μM	89	1.39	4.52	16.7	4.39	30.0	15.2
•0.1μM	192	1.39	4.52	16.7	4.39	20.5	18.0
•1μM	132	2.78	6.02	16.7	4.39	29.5	8.67
•3μM	97	0	0	16.7	4.39	48.6	11.2
•10μM	250	2.78	6.02	16.7	4.39	20.0	9.40
•30μM	133	0	0	16.7	4.39	32.2	7.89
PTZ (day)
•Control	26	0	0	46.2	6.91	15.8	6.43
•2.5 mM	55	0	0	35.7	9.06	42.1	11.4
•5 mM	162	0	0	32.1	8.83	34.3	18.0
•7.5 mM	104	0	0	32.1	8.83	49.9	14.7

A table showing the performance of each classifier. Each classifier sought to separate the data shown in the comparison column, e.g., hcrtr^+/+ (WT) and hcrtr^−/+ (Het). For the pharmacological comparisons, each condition was compared with the rest of the conditions grouped together, aside from the control data which was excluded. For each comparison, 250 motifs were chosen by mRMR, then a smaller number were retained (see motifs column) based on classification error curves (see Extended Data Fig. 4-1A). Cv, 10-fold cross validated; Std, SD across the 10 folds; Mc, majority class classifier; EP, SE of proportion; RM, classifiers built from random motif subsets; WT, hcrtr^+/+; Het, hcrtr^−/+; Hom, hcrtr^−/−.