Skip to main content

Table 2 The varying nature of batch effects in the three datasets as detected by Harman

From: Risk-conscious correction of batch effects: maximising information extraction from high-throughput genomic datasets

PC indices

1

2

3

4

5

6

7

8

A. Correction Vector (Hn-.95)

Dataset 1

0.26

0.33

0.51

0.9

0.44

0.85

0.74

1

Dataset 2

0.42

1

0.93

1

0.99

1

1

0.95

Dataset 3

0.76

1

0.35

0.69

1

1

1

1

B. % of data variance explained by PC

Dataset 1

43.4 %

9.5 %

4.8 %

4.3 %

2.7 %

2.4 %

2.2 %

2.0 %

Dataset 2

19.1 %

11.5 %

6.9 %

4.6 %

4.3 %

4.0 %

3.6 %

3.6 %

Dataset 3

33.9 %

17.2 %

16.0 %

8.6 %

5.8 %

4.5 %

3.7 %

3.3 %

  1. (A) Shows the ‘correction vector’ spanning the first eight principal components for the three datasets resulting from Harman (.95). No or negligible correction were detected for the remaining PCs. A score of 1 means no correction, whereas a score of 0 means maximum correction within the confines of Harman. (B) Shows the relative proportion of overall variance explained by each of the (first eight PCs) for the three datasets
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us