Affymetrix GeneChip Expression Arrays
The Affymetrix GeneChip is the most popular microarray in the world. It is a single-color oligonucleotide array with several important features:
- Rather than use a single long probe, Affymetrix uses a collection of very short probes. The signal from these probes is aggregated into a probeset. This approach has several advantages:
- The short probes increase sensitivity, while the aggregated probeset maintains specificity
- By moving the probes along the gene, one probeset can effectively interrogate regions anywhere from 25 to several thousand base pairs long
- By physically distributing the location of probes on the array, the probeset becomes resistant to localized blemishes or debris.
- Hybridization, staining, washing, and scanning of Affymetrix arrays are performed with specialized robotics, which increases both reproducibility and throughput.
- Primary analysis software and annotation are freely available from Affymetrix
Affymetrix has recently added two new array types to its popular line of expression arrays, leaving many investigators unsure of the best platform for their research.
| 3' | Gene | Exon | |
|---|---|---|---|
| Gene Content | Known and EST, ca 2001 | Known, ca 2006 | Known, EST, and Predicted, ca 2004 |
| Probe Placement | 3' UTR | Full Gene Length | Full Gene Length |
| Probeset Represents... | Gene Transcript | Gene | Exon |
| # Probesets (Hu) | 54 thousand | 29 thousand | 1.4 million |
| Organisms Available | Dozens | Human, Mouse, Rat | Human, Mouse, Rat |
3' Expression Arrays
This is the familiar Affymetrix array design that has been available for several years. Examples include the Human U133 Plus 2.0 and Mouse 430 2.0 arrays. As with most other microarrays, the probes on this array primarily target the 3' end of the gene. This means that the data is essentially an extrapolation - "If the 3' UTR is present at this amount, we assume that the rest of the gene is also present". Because these chips are several years old, some probes are in introns or even outside genes entirely. Even so, the majority of the probes are correctly mapped. A very large amount of data has been run on these 3' Expressions arrays with great success and their accuracy is well-established.
Gene Arrays
The Gene arrays are updates to the 3' design. The design of the array is much newer, so there is less uninformative content. The probes themselves are spread over the length of the gene, so the implicit bias that exists in the 3' arrays is gone. Finally, the array itself is about half the price of the 3' arrays.
Exon Arrays
Exon arrays represent an exciting new direction for gene expression microarrays. Rather than reporting the expression level for a single gene or transcript, exon arrays report the level of every single exon in the genome. This allows for the detection of most varieties of splice variants, known or unknown. Using a process analogous to probeset summarization, we can aggregate the signal of several exons and report a gene-wide signal so that gene abundance can still be assessed. Probesets are assigned to one of three levels of confidence, based on the supporting evidence available:
- Core: Exons that are well documented in the literature - for example, anything in RefSeq
- Extended: Exons for which there is some evidence, but have not been well defined - for example, ESTs
- Full: Exons which have been predicted by a program such as GENSCAN, but which have little to no supporting evidence yet.
When performing analysis on these arrays, it is simple to include or exclude probesets based on the level to which they are assigned. By default, the FGSR uses only the Core-level probesets in our analysis
When planning an exon study, it is very important to include at least three replicates for each condition because alternative splicing detection uses an ANOVA model.
Affymetrix Software
The FGSR recently upgraded to the Affymetrix GeneChip Command Console for all instrument control and data acquisition. This is good news for our users as Affymetrix separated the instrument control and analysis functionality, so you no longer need to download GCOS to view your data.
Primary analysis of all expression data is performed with the Expression Console. Using this software, users may experiment with several different normalization techniques and (for Exon arrays) annotation levels.
For splice variant analysis, the FGSR uses Biotique XRay, a Microsoft Excel plug-in. Results will be given to you in an Excel workbook, which will open without XRay. However, to see the details for each variant found, you'll either need to purchase XRay or use our copy.
When analyzing any Affymetrix array, an invaluable tool is the NetAffx database, which includes sequence information and up-to-date annotation for every Affymetrix microarray.
Affymetrix Data
Raw data will be bundled into a zip file and made available for download. Note that the format of Affymetrix files changed with the new Command Console software, and some older software cannot open these files. If you need the old format files, contact the FGSR and we will convert them for you.
