|
These
comparison tables have been compiled by Y.F.Leung of Department of Ophthalmology
and Visual Sciences, Chinese University of Hong Kong. He has an interesting
web site "yfleung's
functional genomics home" which contains useful information on
functional genomics.
Whilst every care has been taken in the production of this comparison
Technology Networks Ltd., cannot accept liability for any errors or omissions.
Comparison of Microarray
scanners (pdf file)
Comparison
of Manual Arrayers (pdf file)
Comparison
of Automatic Arrayers (pdf file)
Guide to
microarray hardware
- a researcher perspective
Yuk Fai
Leung1, Chi Pui Pang1 and Ken Browne2
1. Department of Ophthalmology & Visual Sciences, The Chinese University
of Hong Kong, Hong Kong.
2. Lab-on-a-Chip.com, Technology Networks Ltd. Crestland House, Bull Lane
Ind. Estate, Acton, Sudbury, CO10 0BD, UK
E-mail:
yfleung@cuhk.edu.hk
cppang@cuhk.edu.hk
kbrowne@TechnologyNetworks.net
Abstract
Well begun
is half done. The selection of appropriate microarray hardware and experimental
strategies are critical for a successful microarray project. In this review
we will not just provide a comprehensive comparison of commercial manual
and robotic arrayers and scanners, but also discuss the hidden cost behind
microarray experiment, the strategies of choosing the right options and
alternatives to buying hardwares to avoid unnecessary expenditures.
~ The mechanic, who wishes to do his work well, must first sharpen
his tools.~
Analects of Confucius 15:9
Researchers
want to perform microarray experiment because it is a state-of-the-art
technology that can increase the chance of publishing better article,
and also deliver tremendous information about the particular system being
studied. A recent survey conducted by The Association of Biomolecular
Resource Facilities (ABRF) microarray research group estimated the mean
cost for setting up a microarray facility is $286,000 (range= $20,000
- $700,000; SD = 162,000).1 The cost includes preparation of cDNA, acquiring
instrumentation, reagents and label targets. This cost is not trivial
for most of the laboratories. As a result, careful consideration before
investing on microarray hardware is necessary. The choice of appropriate
hardware highly depends on the aim of project. In this article we will
provide a brief comparison of the major equipments in microarray experiment,
which is arrayer and scanner, and a purchasing guide for the most suitable
hardware.
Hardware
or not?
Nowadays
some institutes have a central microarray lab that provides most of the
hardware and manufacturing services. However most of them still don't
have such infrastructure. In this scenario, researchers might have a misconception
that they need to purchase both the arrayer and scanner to start their
microarray project. Indeed there are many commercial microarray, arraying
or scanning services that can fulfill this goal. They are most suitable
for those who want to perform only a few experiments and hence don't need
to invest so much money on the hardware. However these services suffer
from two major problems. Firstly it is not economic to perform large-scale
experiment that requires many arrays and replications. Besides, they are
rather inflexible in array design and scanning parameters control. The
result obtained is usually the raw image and extracted data saved in CD-ROM
format. This may not satisfy those researchers who want better quality
control of their data analysis starting from image acquisition. In this
case the researchers should consider purchasing the appropriate hardware.
Arrayer
or not?
Whether
to purchase an arrayer or not is the most critical decision that can affect
the final strategy of microarray experiment. The cost for an arrayer ranged
from a few thousand dollars for a manual type to tens of thousands and
over a hundred thousands dollars for a robotic type of entry and large-scale
level respectively. The top three most widely used arrayer brands in ABRF
2001 survey are Affymetrix/GMS (23%), Pat Brown/ In-House (17%) and GeneMachines
(16%). Purchasing an arrayer is only the beginning of the investment.
There are a lot of hidden costs that researchers have to consider before
making a decision.
Hidden
cost behind arrayer
Printing
an array with tens of thousands of spots requires the same number of clones.
It is not cheap to purchase a clone set. For example, tens of thousands
dollars are required to purchase a commercially distributed IMAGE clone
set.2 Although some of them are claimed sequence verified, there were
reports of high contamination and error rate on the clone set.3-4 The
researchers were suggested to sequence verified themselves. This sequencing
reagent is another substantial hidden cost. Some companies like Clontech,
Incyte Genomics and Operon etc. are supplying array-ready sequence verified
PCR products or designed oligos. (http://ihome.cuhk.edu.hk/~b400559/array.html#Accessaries)
Although it may have better quality assurance, the price is still quite
high (tens of thousands) for producing limited number of microarray. It
is still advisable to maintain a clone set for long-term production.
During the
course of microarray production, numerous PCR reactions have to be done
to amplify enough DNA fragments for printing. Gel electrophoresis of PCR
products is essential for quality control. Given the scale of this PCR
reaction, many PCR machines, liquid handling robots and PCR clean up kits
might be necessary. The recurrent cost for such PCR preparation is also
a factor to be considered.
Together
with the arraying process, these preparations are very time-consuming
and laborious. A few dedicated staff is necessary to keep a smooth microarray
manufacturing process. In the ABRF 2001 survery1, the average number is
4 staff (SD = 0.7). In average about 40% and 10% of the laboratories require
6-12 months and 1-2 years to acquire satisfactory experimental data after
hardware installation. The long-term cost for this extra staff force might
also be a burden to the lab.
Commercial
pre-spotted arrays, Not-for profit arrays
One alternative
would be using the commercial ready-made. There are a large variety of
commercial arrays available from different companies
(http://ihome.cuhk.edu.hk/~b400559/array.html#Microarray slide). However
this is not a long-term solution for large-scale projects because the
commercial arrays are quite expensive. There are several microarray centres
that are distributing microarrays on a not-for-profit basis. For example,
The Ontario Cancer Institute Microarray Center (http://www.uhnres.utoronto.ca/services/microarray/index.html)
is distributing several human, mouse and yeast array products under this
basis. The cost would be low enough even for large-scale studies by using
such products. If the users are willing to sacrifice the flexibility of
choosing the clones to array and the design of the chip, this is definitely
a suitable alternative over spending on the hardware. By using such pre-spotted
products, researchers are still encouraged to purchase their own scanner
for the sake of convenience and data quality.
Commercial
arraying sevice
Commerical
arraying service offers better flexibility than buying the microarray
off-the shelf. The researchers can either choose the clones/ oligos from
the company database or provide the PCR products to the company for arraying.
Again, researchers should be cautious on the hidden costs for sequencing,
PCR and purification process if they are planning to supply the PCR products
to the company for arraying, especially when the number of clones is large.
Which
kind of arrayer?
After considering
the cost factors and alternatives, the users may still find purchasing
an arrayer is the most suitable choice. The researchers are suggested
to refer to several reviews for a comprehensive overview of different
arraying technologies5-9 and array manufacturing procedures5,10, and then
consider for a manual arrayer (Comparison
of Manual Arrayers (pdf file)) or robotic arrayer (Comparison
of Automatic Arrayers (pdf file)) based on the scale of their
project. It may be advisable to start with a cost effective manual arrayer
if there are not many clones and scale up to robotic one if necessary
Besides
the commercial arrayers, researchers may choose to assemble a robotic
arrayer themselves. There is a detailed guide (The MGuide) available from
Prof. Pat Brown's homepage (http://cmgm.stanford.edu/pbrown/mguide/index.html)
on detailed design, construction, parts list and operation of a robotic
arrayer.
Pin (contact)
or Inkjet (non-contact)?
Spotting
pin has become the standard of microarray printing technology nowadays.
All the top three most widely used arrayer brands in ABRF 2001 survey
adopt this technology in their platforms.1 The running cost for using
spotting pins is cheaper than inkjet, the achievable spot size is smaller
and the printing density is higher. The number of spotting pins can easily
be scaled up to achieve higher-throughput. While the inkjet technology
can load much larger volume to the tip for dispensing, the printing volume
is programmable and the print speed per nozzle is much higher than spotting
pins. Which technology to choose is highly depended on the project type
and personal preferences.
Environmental
control?
Some of
the arrayers are equipped with environmental control function like enclosed
cabinet, HEPA filter, humidity control, temperature control and source
plate cooling. These are most suitable for researchers located at a place
with substantial environmental variation. However these measures might
not be mission critical. For example, the Pat Brown type arrayer situated
in Dr. Joseph DeRisi's laboratory in UCSF has been working in open air.
The only environmental control is the central air-conditioning. The absence
of other environmental control has never undermined the quality of his
group's work. (personal communication)
Scanner
or not
Another
piece of hardware is the scanner for acquisition of fluorescent microarray
image (Comparison
of Microarray scanners (pdf file)). It is essentially a fluorescent
microscope that specialized for acquiring microarray fluorescent image
on the standard microscopic slide format. The top three most widely used
scanner brands from ABRF survey 2001 are Packard/ GSI Lumonics (34%),
Axon (24%) and Affymetrix/ GMS (13%). (Figure 3) There are many factors
that determine the performance of a microarray scanner like sensitivity,
resolution, dynamic range and detectivity etc A few comprehensive reviews
on the technology and design of microarray scanner were already published7-8,11-12,
so we are not going to discuss these parameters one-by-one. On the contrary,
we will pinpoint several critical considerations during the decision process.
Commerical
scanning service
Similar
to the situation for arrayer, scanner is not a must for performing microarray
experiment. The alternative would be using microarray scanning service
offered by several companies. Definitely this is not a convenient and
flexible option, since the users need to ship the hybridized chip back-and-forth.
This possesses the risk of photo-bleaching. Besides, researchers cannot
vary the scanning parameters. The choice of optimal parameters is critical
for a successful data analysis. Given the cost of arrayer is not quite
expensive nowadays, it is advisable to purchase one if necessary.
Two lasers
or multiple lasers?
Almost all
of the microarray experiments performed on microscopic slide format used
Cy3 and Cy5 as labels nowadays. There is very little documentation on
trying alternative dyes as labels. Although a recent one that showed Alexa
dye have better signal than Cy313, there will be sometime before these
alternative dye options can be proven to provide the same responses as
Cy dyes and assimilated to current published data. Besides, there is still
no report on comparing the differential expression based on binding more
than two samples simultaneously on the same microarray to binding only
two samples situation. Therefore a scanner with two lasers for Cy3 and
Cy5 labeling is fairly good enough for most of the microarray experiments
at this stage. There are several unique situations that multiple lasers
are necessary: Firstly if the researchers are focused on developing chip-based
SNPs detection, multiple lasers are essential for simultaneous detection
of all four nucleotides polymorphisms. Besides, if users are concerned
on the array production and spotting quality, they can use an extra third
flurophore attached to a sequence that specifically bind to a linker region
of the DNA spots. This can be utilized as a mean to check for the quality
of the arrayed spots. (Neil Winegarden, Ontario Cancer Institute Microarray
Centre, personal communication)
Upgradeable?
Some researchers
may concern about the upgradeability for the scanner. They might feel
uncomfortable if the scanner cannot be upgrade to multiple lasers version.
We believe the same arguments from last section can help to make a sensible
judgment.
Which
technologies to use? (Data comparability across different scanning
technologies)
Perhaps
the most difficult decision to make is which kind of scanning technologies
to choose from when different companies claimed their proprietary technologies
could acquire superior images than their rivals. A better image almost
always means better data and hence a higher chance to have a significant
result. Besides there are raising concerns on microarray data sharing
for cross validation and meta-analysis of combined data from different
laboratories all over the world.14,15 Undoubtedly the reproducibility
of data from the image acquired by different scanner platforms is essential
to make the sharing successful. Therefore it is critical to have a scanner
with good image acquisition and its data is comparable with other commercial
platform.
In general
there are four major technological differences between commercial microarray
scanners: 1. The detection unit is either Charged Coupled Device (CCD)
or Photo-Multiplier Tube (PMT); 2. The image acquisition mode is either
simultaneous or sequential; 3. The optical system is either confocal or
non-confocal. It has been controversial which technology is superior to
its counterpart. A recent report that compared several commercial PMT-based
microarray scanners provided preliminary answer on the last two technological
differences.16 The study demonstrated the fluorescent ratios from scanners
using different technologies were highly correlated. That means users
can basically choose the scanner with simultaneous or sequential scanning
and confocal or non-confocal system that they feel comfortable with and
can still obtain comparable data with others.
There is
limited study to effect using of CCD and PMT on data collection quality.
However based on existing knowledge on CCD and PMT, it is generally believed
PMT can deliver better data than CCD in microarray applications.11-12
Conclusion
The well-chosen
hardware can help to obtain high quality raw data and pave the road to
a good result. However, it can never substitute the personnel who operate
the machines and perform data analysis. Indeed the most challenging topic
nowadays is how to make sense of microarray raw data.17-19 This is not
guaranteed by purchasing the finest hardware. Therefore researchers should
also consider their ability of data analysis or whether they can obtain
help from other groups before investing too much in the hardware.
References
1. Grills,
G. et al. (2001) 2000/2001 ABRF microarray research group study: a current
profile of microarray laboratories. ABRF Microarray Research Group. http://abrf.org/ABRF/ResearchCommittees/MARG/marg.html#Electronic_Posters
2. Bowtell, D.D. (1999) Options available--from start to finish--for obtaining
expression data by microarray. Nat Genet. 21(1 Suppl):25-32
3. Halgren, R.G. et al. (2001) Assessment of clone identity and sequence
fidelity for 1189 IMAGE cDNA clones. Nucleic Acids Res. 29:582-588
4. Knight, J. (2001) When the chips are down. Nature. 410:860-861
5. Eisen, M.B. and Brown, P.O. (1999) DNA arrays for analysis of gene
expression. Methods Enzymol. 303:179-205
6. Rose, D. (2000) Microfluidic technologies and instrumentation for printing
DNA microarrays. In Microarray Biochip Technology (Schena, M., eds), pp.19-38,
Eaton Publishing
7. Mace, M.L. et al. (2000) Novel microarray printing and detection technologies.
In Microarray Biochip Technology (Schena, M., eds), pp.39-64, Eaton Publishing
8. Worly, J. et al. (2000) A systems approach to fabricating and analyzing
DNA microarrays. In Microarray Biochip Technology (Schena, M., eds), pp.65-86,
Eaton Publishing
9. Englert, D. et al. (2000) Production of microarrays on porous substrates
using noncontact piezoelectric dispensing. In Microarray Biochip Technology
(Schena, M., eds), pp.231-246, Eaton Publishing
10. Germino, F.J. and Moskowitz, N.K. (1999) Screening for protein-protein
interactions. Methods Enzymol. 303:422-450.
11. Schermer, M.J. (1999) Confocal scanning microscopy in microarray detection.
In DNA Microarrays: A Practical Approach (Schena, M., eds), pp.17-42.
Oxford University Pres
12. Basarsky, T. (2000) Overview of a microarray scanner: design essentials
for an integrated acquisition and analysis platform. In Microarray Biochip
Technology (Schena, M., eds), pp.265-285, Eaton Publishing
13. Wildsmith, S.E. et al. (2001) Maximization of signal derived from
cDNA microarrays. Biotechniques. 30:202-208
14. Geschwind, D.H. (2001) Sharing gene expression data: an array of options.
Nat Rev Neurosci. 2:435-438
15. Becker. K.G. (2001) The sharing of cDNA microarray data. Nat Rev Neurosci.
2:438-440
16. Ramdas, L. et al. (2001) Comparative evaluation of laser-based microarray
scanners. Biotechniques. 31:546-550
17. Brazma, A. and Vilo, J. (2000) Gene expression data analysis. FEBS
Lett. 480:17-24
18. Quackenbush, J. (2001) Computational genetics computational analysis
of microarray data. Nat Rev Genet. 2:418-427
19. Wu, T.D. (2001) Analysing gene expression data from DNA microarrays
to identify candidate genes. J Pathol. 195: 53-65
|