291using, e.g., a 20% peak threshold and a signal/noise ratio of 3.
Once an appropriate peak threshold and signal/noise ratio is
established this should be held constant for the entire experi-
mental analysis.- The biomarker software applies the statistical techniques as
listed in Table 3. Although the biomarker wizard software
automatically performs the appropriate statistical analysis for
the data set under test, the output obtained does not state
which statistical test has been applied. If the sample size is too
small to test for normality, non-parametric tests are applied. - The plasma proteome is made up of a small number of very
abundant proteins, most notably albumin and IgG and a large
number of low abundance proteins. Considerable concern has
therefore been expressed that these high abundance proteins
can mask the presence of low abundance proteins and efforts
have therefore been made to remove them. Early work used
albumin binding dyes to remove albumin and protein A or
protein G to remove IgG and specialized affinity columns have
15
10
5
0
15
10
5
0
15
10
5
0
5000500010000100002000020000150005000 10000 15000 2000015000Fig. 1 Examples of low molecular weight SELDI-TOF MS spectra obtained from three plasma pools on CM10
chips run at pH 4. Results are displayed using a spectral view (top of figure) and a gel scan view (underneath).
Plasma pools were prepared from xenograft controls (top), xenografts treated with a clinical dose of epirubicin
(middle) and xenografts treated with a maximum tolerated dose of epirubicin (bottom). Difference in peak
intensity can be observed between the three pools
Proteomic Protocol