These two articles rightly emphasize that RNA splicing measures are extremely important and poorly served by array methods so far. However, there are some problems with using exon junctions for array assays as they propose. (1) Many alternatively spliced RNAs involve multiple exons (e.g. Dscam has 130 exons). (2) If you measure expression levels at two different junctions in the same gene, it is hard to tell if they actually occur together on any RNA molecule in the complex mixture; i.e the assay is only local and therefore the encoded proteins are unknown. (3) The choice of probe (critical for getting specificity) is extremely limited for junction probes relative to exon probes (often 100 times more limited).

To address these three problems, we have developed a method to amplify single molecules and ask what exons are found in each molecule in a mixture. A gene with N exons means up to N² possible splice junctions and 2^N possible RNAs. Even if only subsets are used in a given cell, one has to determine which. Our method uses only N exon probes...  Read more

These two articles rightly emphasize that RNA splicing measures are extremely important and poorly served by array methods so far. However, there are some problems with using exon junctions for array assays as they propose. (1) Many alternatively spliced RNAs involve multiple exons (e.g. Dscam has 130 exons). (2) If you measure expression levels at two different junctions in the same gene, it is hard to tell if they actually occur together on any RNA molecule in the complex mixture; i.e the assay is only local and therefore the encoded proteins are unknown. (3) The choice of probe (critical for getting specificity) is extremely limited for junction probes relative to exon probes (often 100 times more limited).

To address these three problems, we have developed a method to amplify single molecules and ask what exons are found in each molecule in a mixture. A gene with N exons means up to N² possible splice junctions and 2^N possible RNAs. Even if only subsets are used in a given cell, one has to determine which. Our method uses only N exon probes instead of N² junction probes to measure the whole set. We are very interested in having this new technologies available broadly. Our first paper heading this direction has been published (pdf of Mitra & Church, 1999), and a couple more are in preparation.

View all comments by George Church