Characteristics and Validation Techniques for PCA-Based Gene-Expression Signatures

Table 1

The measures of coherence, robustness, uniqueness, and transferability applied to all datasets for all signatures used.

Signature	Dataset	Coherence		Robustness		Uniqueness		Transferability
		Signature PC1 (%)	Random mean PC1 (%)	Signature PC1/PC2	Random mean PC1/PC2	Signature correlation	Random mean correlation
Gender-29	DC	45.3	10.9	4.57	1.29	0.048	0.585	0.830
Gender-29	Breast	11.8	12.3	1.32	1.31	0.804	0.606	0.000
TvsN-100	Breast	46.7	9.3	5.28	1.29	0.036	0.815	0.919
TvsN-100	DC	58.5	8.0	9.70	1.27	0.313	0.774	0.975
TvsN-50/50	DC	33.3	8.0	7.09	1.27	0.345	0.779	0.932
TvsN-25/75	DC	18.8	8.0	3.02	1.27	0.385	0.777	0.866
TvsN-10/90	DC	9.4	8.0	1.40	1.27	0.555	0.779	0.634
Mix-29/29	DC	35.2	8.9	1.55	1.29	0.331	0.697	0.008
Mix-29/24	DC	32.1	9.1	1.30	1.28	0.338	0.676	0.003
Mix-29/19	DC	28.5	9.3	1.07	1.28	0.255	0.660	0.225
Mix-29/14	DC	30.7	9.6	1.22	1.29	0.087	0.644	0.598
Gender-29	PRAD	30.8	20.4	1.62	1.68	0.825	0.777	0.025
TvsN-100	PRAD	51.7	16.9	5.72	1.69	0.117	0.949	0.904