Biology Reference
In-Depth Information
approach to circumventing potential artefacts associated with reverse trans-
cription, direct RNA labelling methods have occasionally been employed (
Georg
et al.
, 2009
;
Yu
et al.
, 2011
). A protocol for the generation of strand-specific
cDNA for tiling array hybridization using the FairPlay III Microarray Labeling
Kit (Stratagene) was developed by Roche NimbleGen within the European
BaSysBio project (
Rasmussen
et al.
, 2009
). This method (Protocol 1) involves
incorporation of aminoallyl-dUTP during reverse transcription in the presence of
actinomycin D and subsequent labelling of the cDNA with Cy3. It has been success-
fully applied in a number of recent tiling arrays studies involving
B. subtilis
and
S. aureus
(
Rasmussen
et al.
, 2009; Falord
et al.
, 2011; Buescher
et al.
, 2012;
Nicolas
et al.
, 2012
).
3.3
Array hybridization, scanning and data extraction
In the next steps, the labelled cDNA samples are hybridized to the arrays (usually
overnight), and the slides are then washed and scanned according to the platform
specific protocols and using the respective instruments as discussed in
Section 2
.
For two-colour competitive hybridizations sample and reference, cDNAs (labelled
with Cy5 and Cy3, respectively) are mixed in equal amounts and hybridized to
the same array. As washing artefacts can cause problems by introducing background
variability, the washing procedure is crucial for the quality of the resulting data.
When scanning the arrays, attention needs to be paid to selecting an appropriate
photo-multiplier tube (PMT) settings so that only very few or no probes will be sat-
urated in order to avoid quantitation errors. However, those lower sensitivity settings
have the obvious disadvantage of limiting the ability to reliably detect weak signals.
Importantly, a wider dynamic range is obtained by the extended dynamic range
(XDR) scanning implemented in the Agilent scan control software, thus covering
a larger range of transcript abundance. In XDR mode, the scanner performs a dual
scan (at 100% and 10% PMT, respectively) and the two images are automatically
combined when imported into the Feature Extraction software (Agilent Technolo-
gies; see next paragraph).
Finally, the data are extracted from the scanned image by using appropriate image
analysis software generally supplied with the microarray scanner, but platform-
independent solutions are available as well. Data extraction basically involves grid
alignment, spot finding, rejection of outlier pixels and some kind of background cor-
rection. Additionally, in the case of two-channel arrays, a normalization step is
required, typically involving Loess normalization which corrects for intensity-
dependent dye bias. As a result of the data extraction process, probe intensities
(and ratios) as well as statistical confidence measures associated with each probe
intensity (and ratio) are calculated. When working with the Agilent microarray plat-
form, image analysis is usually performed using Feature Extraction software which
provides a series of spot quality measures and the special feature of summary values
for measurement uncertainty for one-colour arrays (i.e. the processed signal errors)
and two-colour arrays (i.e. the log-ratio errors) based on the selected error model.
