Routine expression analysis of complete mammalian genomes will require the extension of several existing technologies. Spotters will need to be able to print features more densely, and the cost and labor involved in preparing probe DNA for slide manufacture needs to be reduced. To meet these goals, we employ non-contact piezoelectric ink-jet print heads. These print heads can produce features as small as 50 microns, and we routinely spot samples at densities allowing 100,000 features per slide. With 75 micron center-to-center spacing, one would be able to spot 200,000 features per slide. Another advantage of ink-jet print heads is that features can be printed without stopping spotter movement, and can therefore achieve printing speeds of 20 spots per second per printing head. DNA preparation by conventional PCR is labor intensive, time consuming and expensive. One hundred thousand PCR reactions require greater than 1,000 96-well plates, and 100 microliter reactions use nearly $1 worth of enzyme per reaction. This will prohibit many labs from routinely manufacturing complete mammalian chips. We have developed a surface attachment chemistry that allows the probe DNA to be amplified directly on the slide in a single enzymatic process, making the cost and time required to prepare a slide only a few times greater than a single reaction. Using this chemistry, the probes are attached at their 5′ ends to an acrylamide matrix. This approach has two theoretical advantages. First, probe molecules are attached to a volume of acrylamide rather than a two-dimensional surface, which enables a greater amount of probe to be attached per feature. In principle, this could increase the signal-to-noise ratio of hybridization reactions. Second, a 5′ attachment increases the kinetics of target-probe hybridization, and may increase the specificity of hybridization compared with probes attached to the surface at many bases by UV crosslinking. We believe this will allow more stringent wash conditions, and may enhance the specificity of hybridization reactions.
We have also developed the ExpressDB database for yeast RNA expression data and loaded it with 17.5 million pieces of data reported by 11 studies, including data from Affymetrix arrays, cDNA arrays and SAGE. A web-based tool supports queries from 217 conditions (see http://arep.med.harvard.edu/ExpressDB). Clustering analyses of the 217 conditions by similarity of ORF expression profiles indicate that conditions cluster more closely with conditions from the same source study than they do with ostensibly similar conditions from other studies.
This is a preview of subscription content, access via your institution