BMC Bioinformatics

Table 2 Balanced accuracy number of selected features regression models

From: Feature-specific quantile normalization and feature-specific mean–variance normalization deliver robust bi-directional classification and feature selection performance between microarray and RNAseq data

	Breast			Colon
Characteristic	Beta^a	95% CI^b	p-Value	Beta^a	95% CI^b	p-Value
Number of features
10,000	–	–		–	–
5000	0.006	− 0.005, 0.016	0.3	0.003	− 0.007, 0.013	0.5
500	0.025	0.014, 0.036	< 0.001	0.013	0.003, 0.023	0.012
100	0.017	0.006, 0.028	0.002	− 0.003	− 0.014, 0.007	0.5
50	0.012	0.001, 0.022	0.033	− 0.012	− 0.023, − 0.002	0.015
25	− 0.020	− 0.030, − 0.009	< 0.001	− 0.023	− 0.033, − 0.013	< 0.001
10	− 0.053	− 0.064, − 0.042	< 0.001	− 0.054	− 0.064, − 0.044	< 0.001
Model
glmnet	–	–		–	–
SVM	0.001	− 0.005, 0.006	0.8	0.004	− 0.001, 0.010	0.12
Train distribution
Agilent	–	–		–	–
RNAseq	− 0.010	− 0.016, − 0.003	0.003	0.020	0.013, 0.026	< 0.001
Test distribution
Agilent	–	–		–	–
RNAseq	0.007	0.000, 0.014	0.038	0.035	0.029, 0.041	< 0.001
Normalization method
Reference (REF)	–	–		–	–
FSQN	− 0.002	− 0.010, 0.006	0.7	− 0.010	− 0.018, − 0.002	0.010
FSMVN	− 0.002	− 0.011, 0.006	0.6	− 0.013	− 0.021, − 0.006	< 0.001
LOG2	− 0.324	− 0.332, − 0.316	< 0.001	− 0.347	− 0.355, − 0.340	< 0.001

^aBeta Percentage expressed as a decimal, ^bCI Confidence Interval

Back to article page

ISSN: 1471-2105

Contact us

General enquiries: journalsubmissions@springernature.com