Two Stanford University statisticians hope to accelerate the interpretation of genomic data through the use of their Plaid technology, a bioinfor-matics software program that they claim is capable of uncovering relationships among genes that is not possible with other currently available bioinformatics tools.
“The Plaid model is a tool for exploring microarray data in order to discover groups of genes that have similar expression patterns,” said Art Owen, professor of statistics at Stanford. “Plaid allows researchers to uncover relationships among genes that they could never find through correlation, principal components, or hierarchical clustering techniques.”
Plaid technology, developed by Owen and Laura Lazzeroni, exploits genomic information by organizing data into a series of layers, each consisting of a set of genes and a set of conditions or samples. The Plaid algorithm then chooses the set of genes in a layer that have similar expression patterns across the set of samples selected for the same layer. Both genes and samples can belong to more than one layer. For example, gene A can behave like gene B under one set of conditions, but like gene C under another set of conditions.
“For each gene in a layer, the Plaid model estimates the change in the gene’s expression level within the samples belonging to the layer,” said Lazzeroni, assistant professor of biostatistics and genetics at Stanford.
“For each sample in the layer, the model provides an estimate of the up- or down-regulation within that particular sample of the genes belonging to the layer.”
Moreover, Plaid technology is capable of finding more flexible types of relationships and provides results that are easier to interpret than previous methods. For example, there are a variety of options in the search algorithm, including both automated and interactive search strategies, which allow users to specify certain characteristics, such as the number or genes or samples of the layers they wish to find.
“Plaid identifies more flexible types of patterns and arrangements among genes than previous microarray software, and summarizes these patterns in a simple, coherent fashion,” Lazzeroni said. “It tells the researcher not only what genes go together, but when they go together.”
So who will use this new data-mining tool? The Plaid program is designed for scientific researchers who want to search for patterns in their microarray data in the hopes of generating new hypotheses. Presently, the promising technology is receiving some early validation from academia, with more than a dozen universities evaluating the software.
Owen and Lazzeroni hope Plaid technology gains wide acceptance among biotechnology and pharmaceutical firms. The software is available for commercial license through Stanford’s Office of Technology Licensing (OTL), and although OTL has yet to sign up any corporate partners, many firms have expressed an interest.
“I can tell you that we’ve received interest from a number of companies in the biotech area,” says Stefani Shek, a licensing representative with OTL. “We’ve set the licensing fee at $5,000 per year, and we’re hoping Plaid will achieve eight licenses in the next six months.”
Before success comes, though, Owen and Lazzeroni will face challenges common to inventors of new technologies. As the sole entrant in the field of Plaid technology, they will have to create their own markets, and convince the biotech industry, which already has bought into a number of new bioinformatics tools over the past several years, that Plaid technology is a necessary choice. And, while that may seem a daunting task to some, Owen is not worried.
“Plaid technology only has been available for a month,” he said. “I expect a lot of researchers are uncomfortable using the older methods, especially knowing that their rivals could be using a sharper tool.”