Choosing microarray technology Microarrays afford the possibility of measuring gene expression levels for thousands of genes simultaneously. This is accomplished by obtaining RNA from a cell, tissue, or organism, and quantifying the amount of mRNA present for each gene being assayed. One of the most basic considerations when designing a microarray experiment is the type of microarray technology to use. There are two common types of gene expression micro-arrays: (1) ''cDNA arrays,'' usually PCR amplified open reading frames (ORFs) spotted onto glass slides, and (2) oligonucleotide arrays. It is generally accepted that for most applications, oligonucleotide arrays, while more expensive than cDNA arrays, also tend to provide higher specificity (Hudson et al., 2005). Several commercially available oligonucleotide array platforms exist, including Affymetrix GeneChip microarrays, Agilent Oligo microarrays, and Amersham Biosciences (GE Healthcare) CodeLink Bioarrays (Hudson et al., 2005).
The most commonly utilized microarray platforms employ either a two-color or one-color design. Two-color experiments involve co-hybridization of cDNAs derived from two different RNA samples on the same slide. Each set of cDNAs is labeled with a different fluorescent dye (e.g., Cy3 and Cy5). For each gene being assayed the relative expression level in sample #1 versus sample #2 is obtained by the ratio of Cy3 fluorescence to Cy5 fluorescence. One-color platforms, such as Affy-metrix ("Affy" chips), are simpler, in that only one labeled cDNA is hybridized per chip. For each experimental sample, a fluorescence value is associated with each gene present on the microarray. This value can then be compared against the fluorescence value for that gene obtained with a common reference sample, to generate a ratio value analogous to that of a 2-color design.
Replication, variation, and validation A primary criticism against many aging-related micro-array studies is a lack of sufficient replication (Melov and Hubbard, 2004). While the number of replicates required for interpretation of microarray data will vary based on many different factors, including the type of experimental design and the organism under study, there are standard approaches that can help determine the degree of replication needed to obtain statistically meaningful gene expression changes (Hudson et al., 2005). Of particular import when considering replication is the difference between biological and technical replicates. Technical replicates are replicates derived from the same biological sample (e.g., the same pool of RNA), and can refer either to the same probe spotted at different places on a microarray slide or to parallel hybridization of the same sample to multiple slides. In either case, technical replication can provide information only about the error in the measurement and gives no indication of the variation within the population from which the samples originated. Biological replication refers to analysis of samples obtained from different individuals or pools of individuals. When considering the number of independent replicates necessary for statistical analysis, only biological replicates should be counted.
A higher number of replicates may be necessary for aging-related microarray studies relative to microarray studies that examine other aspects of biology. Unlike gene expression changes associated with cancer, for example, which can be several-fold in magnitude for the most up- or down-regulated transcripts, age-associated changes tend to be relatively small in magnitude and can easily be masked by variation between individuals (Hudson et al., 2005). In addition, there is evidence that gene-specific variation increases with age, perhaps due to dysregula-tion of transcription in older individuals. Thus, the most relevant transcriptional changes in an age-related study may be smaller in magnitude and require a greater number of replicates to be statistically detectable.
The first look a researcher usually gets at the data from a microarray experiment is in the form of an image file, where each "spot" corresponds to the fluorescence intensity associated with the expression of a particular gene. The process of getting from the raw data to a list of genes that are up- or down-regulated involves several steps, each of which can have a substantial impact on the quality of the data set. The image must first be processed, which involves manual inspection of the image files for anomalies, followed by spot quantification, quality control, and normalization. Many software packages are available to assist with visual inspection and spot quantification, and the appropriate choice will depend of several factors, including platform and cost (Hudson et al., 2005).
In the early days of microarray use, researchers would often carry out experiments with two or three replicates (sometimes only one), apply minimal normalization procedures, calculate the average fold-change in mRNA levels for each gene, select an arbitrary fold-change cutoff (usually 2-fold), and report a list of genes that are "significantly" up- or down-regulated (Melov and Hubbard, 2004). Fortunately, those days are (mostly) gone. Although there are no universally accepted standards for processing, analyzing, and presenting micro-array data, it is now expected that rigorous statistical methods be applied to any microarray data prior to publication. Several detailed reviews are available on statistical methods for analyzing microarray data
(Churchill, 2002; Kerr and Churchill, 2001; Kerr et al, 2000; Lee and Whitmore, 2002). In most cases a statistician should be consulted both during the design phase of the microarray experiment and during the data analysis phase.
A well-designed microarray experiment will result in a set of genes that, following appropriate statistical analysis, show statistically significant changes in mRNA level as a function of the experimental condition under study. Unfortunately, many studies go no further, in terms of validation, than to list the set of genes that meet this description. In any microarray experiment, independent validation of the observed changes in mRNA levels, or at least a subset of the most important observed mRNA changes, is essential. At a minimum, real-time quantitative PCR (RT-PCR) or some other method should be used to demonstrate the expected change in mRNA level. If biological relevance of a particular gene expression change is to be inferred, then a change in protein level, or even better protein activity, should be shown. Since translational or post-translational regulation can offset changes in mRNA level, there is no guarantee that the activity of the encoded protein has been altered even if mRNA levels are dramatically changed.
Was this article helpful?
Your heart pumps blood throughout your body using a network of tubing called arteries and capillaries which return the blood back to your heart via your veins. Blood pressure is the force of the blood pushing against the walls of your arteries as your heart beats.Learn more...