Biomedical Engineering Reference
In-Depth Information
the type of array and the purpose of the experiment. The most commonly
used are DNA microarrays where oligonucleotides are arrayed to test for a
complementary nucleic acid in the sample applied to the array. The experi-
menter could be looking at RNA from gene expression, SNP and DNA for
genotyping, microRNA, or anything that binds/interacts to a specifi c oligo-
nucleotide sequence. Other arrays focus on different interactions such as
protein-protein, protein-antibody, protein-chemical and biomaterial-cell
interactions. Each of these arrays will require a different method of analysis,
from the experimental design to the meta-analysis of the results. In all cases,
regardless of the nature of the data obtained and in order to be meaningful
to the end user, the same analysis challenges will be encountered: exper-
imental design, normalization, analysis, multiple testing correction, and
interpretation of the results. The trend for miniaturization and multiplexing
assays is not going away. All the analysis challenges due to miniaturization
will be current and no one analysis method will fi t every situation. There
have been more than 1,000 papers published on analysis of microarray data.
Seldom are two papers by different authors to be found using exactly the
same method of data analysis, from signal extraction to the meta-analysis.
As technology develops and becomes more and more common, it is hoped
that there will be a move towards the standardization of the analysis meth-
ods with sound statistical reasoning, interpretation and communication. This
chapter is designed to guide the reader through the many choices available
for each data analysis step.
8.2 What do we measure?
Because of the different types of microarray substrates and differences in
chemical-physical interactions, what is measured is going to vary greatly.
In all cases, what is measured is a signal per assay on the array. The signal
could be an absolute intensity or a ratio of signal intensities (if we have mul-
tiple measurements on the individual assay). One of the most commonly
used arrays measures messenger RNA (mRNA) levels in a cell's response
to perturbation or disease state. The mRNA levels refl ect the genes that
are turned on or off or fi nely tuned in response to a stimulus. The same
concept also holds for microRNAs, although their action is to silence or
tune the expression of groups of genes. Another very popular array is a
genotyping array that measures single nucleotide polymorphisms (SNPs),
that is nucleotides changes a person has in her DNA and correlating that
to disease, disease risks, drug interaction and drug effi cacy. Proteins micro-
array can identify protein-nucleotide or protein-protein interactions.
There are also biomaterial arrays where proteins or whole cells interact
with the biomaterial (Pernagallo
et al.,
2008; Thissen
et al.,
2006), as well as
microarrays that report at the change in gene expression after exposure to
