Biology Reference
In-Depth Information
Quantitative Noise Analysis
G. A. Held, Gustavo Stolovitzky, & Yuhai Tu
The last decade has witnessed a shift in molecular biology from
methods that probe hypotheses a few molecules at a time toward
whole genome measurements. Global gene expression assays have
enabled the monitoring of the transcription levels of tens of thousands
of genes simultaneously [1-3]. In the near future, it will be possible to
profile all of the nonrepetitive sequences in the genomes of higher
organisms [4], including
H. sapiens
, with only a few DNA gene chips.
This will allow one to obtain a global view of the transcriptomes corre-
sponding to different cell phenotypes. Such capability will greatly
accelerate and perhaps fundamentally change biomedical research and
development in many areas, ranging from developing advanced diag-
nostics to unraveling complex biological pathways and networks, to
eventually facilitating individual-based medicine [1,2].
Interestingly, the power of global gene expression assays brings along
its own drawbacks—useful information is typically measured amidst
high levels of noise. In general, the changes in the measured transcript
values between different experiments are due to both biological varia-
tions (corresponding to real differences between different cell types and
tissues) and experimental noise. To correctly interpret this data, it is
crucial to understand the sources of the experimental noise. As will be
demonstrated, systematic study of the noise in such data enables one to
assign a meaningful statistical significance to observed transcriptional
changes. In addition, understanding the sources of noise provides a
useful guide in attempting to improve the technologies. Finally, it pro-
vides an effective and systematic means of comparing the reliability of
the various global gene expression assays available today.
The most mature global gene expression technology is arguably the
microarray. In all of its implementations (cDNA arrays [2], oligonu-
cleotide arrays [1,5], etc.), this transcription profiling method exhibits
significant technology-dependent noise. Models that characterize this
noise through the study of replicate measurements [6] have been devel-
oped and provide a measure of security against false discoveries.
An alternative gene expression profiling method employs the sequenc-
ing of short sequence tags derived from the ends of messenger RNA.
103
Search WWH ::
Custom Search