Skip to main content

Table 2 Number of transcripts with non-zero blending error variance

From: On the relevance of technical variation due to building pools in microarray experiments

Number or

Data set

proportion of

Simulated

 

Experimental

transcripts

s1

s2

s3

 

Mouse

Rat

Bee

Human

Total

9000

9000

9000

 

18646

15923

14400

43256

Crit. >8

9000

9000

9000

 

8554

6264

13761

12348

Sign VC

1

1794

6704

 

4329

6

7093

0

\(\hat {\pi }_{1}\)

0.005

0.295

0.918

 

0.75

0.29

0.68

0.40

  1. Results of the restricted likelihood ratio tests of the hypothesis \(\mathbf {H}_{0}:{\sigma _{2}^{2}}=0\) for transcripts exceeding the minimum expression level (crit. >8). Numbers of transcripts with a significant variance component for imperfect blending (sign VC) were counted according to the FDR correction level of 5 %. \(\hat {\pi }_{1}\) is the estimated proportion of transcripts with \({\sigma _{2}^{2}}>0\). Simulated data sets s1, s2 and s3 refer to scenarios where none, one third, and all transcripts were associated with a non-zero blending error variance component