Microarray-based infrastructure within the BPFG
The BPFG has one of the most functionally flexible microarray infrastructures for small and medium sized microarrays, specifically suitable for project-directed and microbial applications in the UK (or most other places for that matter). We have experience working with PCR-product probes, linkered and un-linkered oligonucleotide probes, and protein microarrays that are produced and available in-house, and also using externally sourced microarrays. These studies span bacterial systems, plant and animal microarrays, and microRNA studies. The forthcoming addition of a high-throughput non-contact printing solution in 2007 will enable us to produce higher density protein and nucleic acid probe microarrays, at which point we hope to be able to produce additional microarrays including complete mouse and human probe sets.
The microarray-directed tools in the BPFG include:
Microarray printers:
Two Genetix Q-Arraymini microarray printers. These printers print up to 54 slides using up to 48 pins. They are very functionally flexible, we have four different types of pins available, and the printers have the full 'protein arrayer' configuration, for maximum flexibility for different applications. These printers have been the basis for most of our microarray printing / production to-date. They are generally very functionally flexible, reasonably quick, and are pretty reliable. They are particularly good for project-directed and microbial sized microarrays associated with our usual applications to-date, but they are not really intended / suitable for high density microarrays.
A Perkin Elmer Piezorray. This is a piezo electric non-contact printer, primarily used for protein array work. It has four tips and prints spots with an approximate delivery volume of 300 picolitres. The advantages of this system are that it's delivery reproducibility is very high, and much higher than that associated with contact printing. When applicable, it is our printer of choice for protein microarrays. The main drawback of this system is that it is comparatively slow, printing 384 different probes to 25 slides in approximately 5 hours. As such it is not really well suited to most nucleic acid probe experiments that we perform. However, when working with probe sets of up to 384 probes, the high quality of the microarrays frequently make this the printer choice, regardless of the nature of the probes. As a general point, non-contact printing requires greater user input, and is technically a little more challenging than contact printing.
Coming soon: An ArrayJet Aj120 microarray printer. This microarray is a higher-throughput non-contact microarray printer, based upon inkjet technology. This system has 16 and 32 nozzle printing heads, and is substantially faster than the other printers we have. It also has a plate stacking facility. This will enable us to produce higher density microarrays using both proteins and nucleic acid probes (and other things too). This printer is the basis of our planned expansion into the production of complete human and mouse probe sets. We also intend to explore using it to increase within-slide replication to enhance data quality with our existing probe sets.
Microarray processing equipment:
Most of our protocols include on-slide mixing, for which we have two 4-position Advalytix SlideBoosters. These use acoustic wave transducers below the microarray to continuously agitate the solutions on the upper surface of the slide. We use these as our method of on-slide mixing because they are very flexible in the size and location of the printed area of arrayed materials that they can be used with, and normally use these with LifterSlips.
Microarray scanners:
A Genepix4000B microarray scanner. This is a single slide, two-colour scanner, which we mainly use only as a back-up scanner. This scanner was 'inherited' from another group, and has the singular advantage that it is extremely reliable. As we see them, it's limitations are that it can only do one slide at a time, the laser power is a little higher than we prefer to use, the raw image resolution is usually higher than we like to use, and the absence of autocalibration functions increase the user-generated variability between replicate scans.
A Perkin Elmer ScanarrayExpressHT microarray scanner. This is a high-throughput scanner, with a 20-slide autoloading system, which we have configured with four lasers (red, yellow, green, blue). This is our 'workhorse' scanner. We have had some slide loading and other set-up problems with this machine, but it is our scanner of choice. The autoloading capacity enables us to do extensive QC scanning of printed microarrays, and to be able to include pre- and post-prehybridization scanning in our protocols. The autocalibration functions are invaluable, especially since once a substantial number of arrays are being worked with the time taken to adjust and optimize scans can be considerable.
A Genicon (when we bought it Qiagen) resonance light scatter microarray scanner. The scanner was originally obtained in order to increase the sensitivity of microarray assays both with and without amplification (preferably without). That experimental need was subsequently overcome using a different strategy through the use of different fluorescent labelling approaches. However, we continue to see useful applications, primarily in the area of protein microarrays, using this technology. The biggest drawbacks, at least from our perspective, of this system is that it suffers from an almost complete lack of technical support, and the scanning control software is not designed to be able to take fully optimal dual colour image datasets. As such, we consider that a good piece of equipment is compromised by inadequate software. The movement of this equipment from Qiagen, to Invitrogen, and now to ?, has not been helpful.
A Perkin Elmer ProScanArray microarray scanner. This is the current replacement, with similar specifications (this one only two-colour) as our ScanArrayExpress HT, and it also has 20-slide autoloading and autocalibration facilities. This scanner was obtained on the basis of our good experiences with the other scanner. We look forward to when it will work as well as we are / were expecting.
Image analysis:
Image analysis has to be separated from data analysis, and is of critical importance to good quality microarray studies. We have three dedicated workstations for image analysis, for which we primarily use BlueFuse (BlueGnome, Cambridge). We have tried many approaches previously, including GenePix versions 3 through 5, and ScanArray, but for our purposes and throughput we consider BlueFuse to be the best option of which we are currently aware and have worked with.
Data analysis:
Our primary LIMS and data analysis tool is BASE, which is maintained and supported by the Dunn School / WIMM Computational Biology Research Group (CBRG), primarily with Dr Stephen Taylor. We like this tool and have found it to be very reliable, especially when working with large datasets and numbers of experiments. We also like the fact that it is clear, simple, and it is easy to track what has been done to the data. We have some additional plug-ins developed with CBRG, for example implementing the Cyber-T Bayesian statistical analysis method, and also to assess microarray consistency and confidence intervals. These plug ins can be obtained through contacting Dr Saunders.
With the CBRG, primarily with Dr Simon McGowan, we have also worked on using interactive GBrowse databases for the presentation of microarray data in its genomic context. Examples of these can be seen by following the links for published datasets from the 'data' pages of the CBRG.