Statistical techniques to construct assays for identifying likely responders to a treatment under evaluation from cell line genomic data

Table 2 Predicted dacetuzumab sensitivity versus GCB/ABC subtype

Classifier	Odds Ratio/95% CI		GCB	ABC	Unknown
Lasso/RF	∞	Sens	0.824	0	0.176
		Semi	0.655	0.167	0.179
		Resis	0.359	0.427	0.214
SNSS/SNSS	208 (25.40, 1703.23)	Sens	0.889	0.011	0.100
		Semi	0.361	0.373	0.265
		Resis	0.208	0.542	0.250
SNSS/RF	81.71 (10.50, 635.96)	Sens	0.898	0.020	0.082
		Semi	0.625	0.205	0.170
		Resis	0.250	0.464	0.286
SNSS/KNN	29.03 (6.53, 129.17)	Sens	0.769	0.031	0.200
		Semi	0.591	0.303	0.106
		Resis	0.344	0.400	0.256
Lasso/KNN	6.38 (2.69, 15.16)	Sens	0.613	0.132	0.255
		Semi	0.600	0.338	0.062
		Resis	0.320	0.440	0.240
RDA/RF	4.31 (1.91, 9.72)	Sens	0.657	0.157	0.186
		Semi	0.612	0.194	0.194
		Resis	0.393	0.405	0.202
RDA/KNN	2.31 (0.95, 5.61)	Sens	0.661	0.161	0.179
		Semi	0.477	0.295	0.227
		Resis	0.532	0.299	0.169

The odds ratio for GCB subtype by dacetuzumab sensitive versus resistant classification, with a 95% confidence interval. The GCB/ABC/Unknown subtype distribution is shown also for the semi-sensitive class, this class is excluded from the odds ratio calculation for simplicity. Two of the classifiers, Lasso/Lasso and RDA/RDA, selected no genes and have therefore been excluded from this table.

ISSN: 1471-2407