ing=0 width=100%>Yahoo! is not affiliated with the authors of this page or responsible for its content.
Choosing Components for a Microarray Scanner Choosing Components for a Microarray Scanner
Page 1
Choosing Components for a Microarray Scanner
Stephanie Weiss, For Hamamatsu Corporation
The Human Genome Project is producing a
huge volume of data about the structure,
organization and function of the estimated
50,000-100,000 genes within our DNA.
Understanding their function, in particular,
could improve ability to diagnose, treat and/or
prevent disease.
Given the volume of interesting genes,
the best way to study their function is with
massively parallel analysis, e.g., gene
expression microarrays.
Gene expression occurs when genetic
information contained within DNA is
transcripted into messenger RNA (mRNA)
molecules that are then translated into the
proteins that perform critical cell functions.
Changes in the types and amounts of mRNA in
a cell can indicate how the cell responds to
environmental stimuli or other changes.
To study mRNA, researchers exploit the
fact that it will bind specifically (hybridize) to
the DNA template of its origin. By combining
fluorescent markers with the mRNA, scientists
can use photonics to quantify the amount of
mRNA that binds to a specific DNA sample. By
placing many DNA samples on one microarray,
scientists can study, in parallel, the expression
levels of hundreds or thousands of genes within
a cell.
The usefulness and repeatability of this
analysis depends on a large part on the photonic
components and technologies lasers, detectors
and optics used in reading the microarrays.
How microarrays work
A microarray is a small surface for
example, a microscope slide onto which a
researcher can place or synthesize many
hundreds or thousands of tiny samples of DNA,
cDNA, or oligonucleotides (fragments of single-
stranded DNA).
After
immobilizing the DNA target on
the array surface
, the researcher labels it with a
fluorescent probe and allows it to hybridize for
two to 12 hours. A gene scanner then detects the
amount of labeled probe that hybridizes to the
DNA; the intensity of the fluorescent light
varies with the strength of the hybridization.
For some applications, such as
genotyping, binary detection (fluorescence or no
fluorescence) may be adequate to produce a
result.
However, gene expression and single-
nucleotide polymorphism (SNP) studies
quantify differences in intensity. For this reason,
sample preparation, microarray surface
uniformity and gene scanner repeatability are
critical.
Microarrays usually use probes labeled
with two fluorophores, commonly cyanine 3
(Cy3, with peak absorption at 550 nm and
emission at 570 nm) and cyanine 5 (Cy5, with
peak absorption at 649 nm and emission at 670
nm).
To analyze the slide, a microarray reader
uses a light source a laser or lamp to excite
the fluorophore(s). A photomultiplier tube or
CCD camera then detects the resulting
fluorescence, and the system produces an image
that shows the intensity ratio between the two
fluorophores, generally reported as Cy5:Cy3.
Conventionally, yellow on a ratio image
indicates a 1.0 (1:1) expression ratio (no
difference in expression between samples). Red
is higher expression levels of the Cy5 sample;
green is higher levels in Cy3 sample. Black
(dark) spots indicate that neither sample
expressed. Choosing Components for a Microarray Scanner
Page 2
Diagram 1, Scanning Systems.
Reading the slide
There are two basic types of microarray
readers:
Scanning systems use narrowband
illumination (i.e., lasers) to excite the
fluorophores, then capture the resulting
fluorescence with photomultiplier tube (PMT)
detectors. See Diagram 1, above.
The PMT converts incident photons into
electrons via the photoelectric effect: a photon
strikes the active surface of the PMT (the
photocathode), generating an electron. The
electron flows through a series of dynodes that
multiply the electrons until they reach the
anode. The resulting current from the anode is
directly proportional to the incident light at the
photocathode.
Staring systems generally use
wideband illumination, such as a xenon lamp, to
excite the fluorophores, then capture the
resulting fluorescence with an array detector,
such as a charge-coupled device (CCD). See
Diagram 2, right.
System design will determine which of
these types of systems can analyze a given
microchip more quickly. Typically, scanning
systems deliver more excitation photons to the
sample, resulting in generation and collection of
more emission photons per pixel in a given
amount of time. Although the staring system can
captures a large portion of a microarray, the
array detectors used are much less sensitive than
PMTs, and the wideband excitation source is
less efficient, so the camera must integrate for a
longer period to capture the same amount of
fluorescence signal as a scanning system.
A scanning system can excite/detect one
color at a time (sequential scanning), or acquire
both at once (simultaneous scanning). The latter
is faster, but improperly designed systems can
suffer from crosstalk between channels:
emissions from (generally) the shorter
wavelength fluorophore creeping into that of the
longer wavelength and resulting in higher-than-
Diagram 2, Staring Systems
. Choosing Components for a Microarray Scanner
Page 3
real fluorescence reading in the longer-
wavelength channel and a corresponding
increase in the ratio.
To avoid crosstalk in a simultaneous
scanning system, some system designers use
lasers tuned to suboptimal excitation
wavelengths that increase spectral separation
between Cy3 and Cy5. Other designers have
developed optical designs and filters to ensure
that only photons at desired wavelengths reach
the detection system.
Choosing components
System designers and users have
determined some standard requirements for
microarray reader technology,

[1]
but
maximizing the signal-to-noise ratio (S/N) is the
ultimate goal because this metric determines the
confidence in the accuracy of a given signal
measurement/the likelihood that a given
fluorescence signal will be visible above the
noise of the system.
System noise comprises several
components:
Background noise, which can come
from intrinsic fluorescence from the glass
microarray substrate, out-of-focus background
signal (stray light and scattered light) and
nonspecific hybridization.
Dark current, which measures the
number of electrons per second that a photon
detector introduces from internal thermal
emissions or leakage current from the dynodes
of a PMT. To minimize dark current noise,
choose a detector, such as the R6358, R4632,
R6060 series, with very low dark current levels
.
Also, illuminate each pixel and integrate for as
short a time as possible. For a CCD-based
system, low-temperature operation is critical for
longer integration times.
Electronic noise from poorly designed
components can also increase dark current.
Detector modules, such as the HC120 Series,,
which include carefully engineered low-noise
electronic circuits, can minimize this type of
noise and reduce system design time.
Shot noise, a random quantum effect.
Because of its relationship with the signal
photons, this type of noise increases with signal
intensity, but only as the square root of the
signal, so the signal-to-noise ratio increases with
signal intensity. Optimal systems are shot noise-
limited because quantum physics says shot noise
will always be there. Shot noise will produce
some difference in intensity between two
identical spots, but printing and hybridization
inhomogeneity will result in much more
significant variations.
Detector choices
To maximize S/N, the first step is to
choose a detector with high sensitivity in the
emission wavelength region of the chosen
fluorophores. [R7400,R3896,H7422] Also, look
for reliability in performance: linearity of output
over a wide variety of incident light intensities.
For a scanning system that uses a PMT,
the PMTs amplification depends on the number
of dynodes in the PMT and the voltage applied.
Gains of 10
7
are possible.
For optimal performance, the user will
want to set the PMT gain so that the brightest
signals use most of the systems dynamic range.
Generally, increasing PMT voltage beyond this
optimal setting does not improve S/N because at
high gain levels noise increases more than
signal. Dropping the PMT voltage below an
optimal range to reduce gain (for example, if the
fluorescence intensity of the sample is saturating
the detector) also does not improve S/N because
the PMTs photon-to-electron conversion
process is not as efficient at low gain levels.
Instead, reduce the laser power.
For a CCD, spatial resolution will be a
significant consideration. To capture a
microscope slide (25 × 76 mm) at 10-µm
resolution, a CCD would need 2500 × 7000
pixels, or two frames at 1600 × 1200, stitched
together. Acquiring and electronically stitching
together multiple images could impact total
analysis time.
Optical considerations
A standard microscope objective has a
high numerical aperture (light collection
efficiency), but its field of view is limited and
not typically very uniform. This type of lens
collects more light in the center than at the