Table 4 Parameter estimates in the ANOVA model. This is a mixed-effects model, as the first three effects are fixed and the others are random. The noise in these experiments was largely due to gene and to the interaction between replicates and gene. Reduction in samples size yielded increased noise since

μ

Fixed overall level

-0.13

C

Sample size; 1000 or 250 cells

0.055

D

Dye ratio; cy3/cy5 or cy5/cy3

0.10

Random effects E~N(0, σ²_E)

Explanation

Estimated standard deviation ()

A

Array; 1,...,12

0.14

G

Gene; 1,...10643

0.47

CG

Interaction: cell size sample and gene

0.044

BG

Interaction: replicate (10 000 cells) and gene

0.17

B₁G

Interaction: replicate (1000 cells) and gene

0.41

B₂G

Interaction: replicate (250 cells) and gene

0.45

DG

Interaction: dye ratio and gene

0.18

ε

Model and measurement error

0.33