Biology Reference
In-Depth Information
between two samples, as variations in spot size or amount of probe on the array do not
affect the signal ratio. However, as the analysis is always based on direct compari-
sons between the co-hybridized samples, the experimental design can be quite com-
plicated and inflexible for large sample sets, which is especially true for interwoven
loop designs (
Churchill, 2002; Khanin and Wit, 2005
). In the case of
in situ
synthe-
sized arrays, the low variability between individual slides is associated with an
improved reproducibility of the data at the signal level thus allowing for one-colour
hybridizations (
Patterson
et al.
, 2006
). Unlike the Affymetrix system, which is solely
based on single-channel detection, the NimbleGen and Agilent platforms allow for
both one- and two-colour approaches. Two-colour, common reference designs can
be an advantage with respect to controlling the variation associated with the labelling
and hybridization procedures but, on the other hand, do not allow for direct compar-
isons of data from different studies. The common reference approach using a pool of
all samples included in a study will be discussed in
Section 2.2
.
Because of the limitations of spotted arrays, such as higher slide-to-slide varia-
tion, lower probe density and less flexibility with regard to probe content, only
in situ
synthesized arrays will be considered in this chapter. However, this does not preclude
the use of spotted arrays for generating reliable data if the technical performance of
array production and processing is sufficiently well controlled and adequate care is
taken in experimental design and data analysis. Two recent examples of such studies
have been published by
Hahne
et al.
(2010)
and
Blom
et al.
(2011)
.
2.1.2
Array designs and array processing requirements
The largest collection of bacterial expression microarray designs (so-called cata-
logue arrays) is offered by Roche NimbleGen, where designs for over 200 bacterial
species are currently available, but will most probably cease to be produced by the
end of 2012. For
Escherichia coli
, a tiling array design is also offered by Roche Nim-
bleGen containing 385,000 probes and having a median probe spacing of 24 nt. In
addition, expression microarrays for many bacterial species are available from
MYcroarray, a company employing maskless photolithography, or Genotypic that
uses Agilent's SurePrint
in situ
technology. Agilent's collection of model organism
gene expression microarrays includes an
E. coli
array containing 15,000 probes, and
Affymetrix's previous generation GeneChip arrays can be obtained as pre-designed
bacterial arrays for
E. coli
,
B. subtilis
, and
Staphylococcus aureus
. The approach
applied by most researchers, particularly with respect to genomic tiling arrays, is
to perform the probe design for the studied organism and then have the arrays pro-
duced by a commercial manufacturer (so-called custom-designed arrays). In contrast
to spotted arrays,
in situ
synthesized, high-density arrays depend on commercial
manufacturing because of the specific production processes employed.
There are many aspects of the probe design that have to be taken into consider-
ation in order to generate high-quality experimental data. Importantly, the thermo-
dynamic properties of all probes on the array should be as similar as possible. Probe
quality parameters include melting temperature (adjustment of probe length for
obtaining a near-isothermal design, i.e. homogeneous
T
m
which correlates to probe
