Global Gene Expression Assays: Quantitative Noise Analysis - Systems Biology: Networks, Models, and Applications - page 132

factors should be noted. First, at higher concentrations, the Affymetrix
signal strength begins to saturate while in principle the MPSS tpm
values should not. Thus, at these higher concentrations, the MPSS data
will present a more accurate measure of fold changes. In addition,
the MPSS data is obtained without any previous assumptions as to
which signatures (and thus which genes) may be present. In contrast,
Affymetrix GeneChips only measure the expression levels of those
genes for which probe sequences are explicitly incorporated onto
the chip. Finally, the MPSS measurements should not be susceptible to
cross-hybridization effects that can affect Affymetrix data for experi-
ments that are less well controlled than the spike-in data addressed
above. That is, replicate noise measurements notwithstanding, each of
these technologies has attributes not present in the other.
SUMMARY
A review has been presented of the means by which sets of biologically
equivalent replicate gene expression assays, which bifurcate at differ-
ent stages of sample processing and data collection, can be used to
quantify the sources of noise introduced into these assays. This tech-
nique has been utilized to carry out detailed analyses of the noise
present in two popular and quite distinct global gene expression
assays: Affymetrix GeneChips and Massively Parallel Signature
Sequencing (MPSS). In the course of these analyses, the differences
between digital and analog expression assays have been illustrated.
A general procedure by which similar analyses may be carried out
for additional global gene expression assay technologies has also
been presented. Finally, it has been shown that the quantitative noise
measurements obtained from replicate experiments allows for a com-
parison of the relative sensitivities of experiments carried out using
different technologies.
ACKNOWLEDGMENTS
The results presented in this chapter were largely derived from refs. [12]
and [13]. The contributions of the following collaborators in these works are
gratefully acknowledged: U. Klein, A. Kundaje, K. Duggar, C. Haudenschild,
D. Zhou. T. Vasicek, K. Smith, A. Aderem, and J. Roach.
REFERENCES
1. Lockhart, D. J. and E. A. Winzeler. Genomics, gene expression and DNA
arrays. Nature , 405(6788):827-36, 2000.
2. Brown, P. O. and D. Botstein. Exploring the new world of the genome with
DNA microarrays. Nature Genetics , 21(Suppl 1):33-7, 1999.