Study Design in DIGE-Based Biomarker Discovery - Difference Gel Electrophoresis (DIGE): Methods and Protocols - page 200

Table 2
Sample size of each group to achieve a power of 0.8 or 0.9 for different effect sizes
and proportions of effective proteins

Sample size of each group

Power = 0.80 Power = 0.90

Effect size p 1 = 0.01 p 1 = 0.05 p 1 = 0.10 p 1 = 0.01 p 1 = 0.05 p 1 = 0.10

1.5 21 16 15 24 19 17

2.0 13 11 9 15 12 11

2.5 10 8 7 11 9 8

3.0 8 6 6 9 7 6

3. Methods

As described in the introduction, it is advisable to subdivide a
biomarker discovery study into two experiments: a “discovery
experiment” for detection of biomarker candidates followed by a
“validation experiment” for their confi rmation.

3.1. Discovery
Experiment

To make an adequate sample size calculation (i.e., to estimate how
many samples have to be collected for the discovery experiment),
the discussion of the following parameters is required: false discov-
ery rate, statistical power, proportion of effective proteins, and
minimal effect size.

3.1.1. Sample Size
Calculation

The FDR value is usually set to 0.05 (see Note 1).

Fix a Level for the False
Discovery Rate

The minimal level for the statistical power which is required for a
discovery experiment is usually 0.8. The statistical power is 1 minus
the type 2 error ( b , i.e., a power of 0.8 corresponds to a type 2
error of 0.2). The type 2 error is the proportion of false-negative
decisions among the assumed number of effective proteins ( m 1 ),
i.e., the proportion of effective proteins not identifi ed using a sta-
tistical test. A power of 0.8 is suffi cient for most genuine biomarker
studies. However, it has to be stressed that such a level may be too
low to draw mechanistic conclusions (e.g., on pathways) from the
results of the discovery experiment (see Note 2).

Fix a Level for the
Statistical Power

p 1 is the ratio of the number of effective proteins ( m 1 , i.e., those
proteins which, in truth, differ between the two experimental
groups) to the total number of proteins on the gel ( m ). The absolute

Make an Assumption
on the Proportion of
Effective Proteins ( p 1 )