Scimagix thinks theres a good chance your company has an image problem.
The Redwood Shores, Calif.-based company estimates that over 70 percent of all data generated in pharmaceutical and life science research is in the form of images, from 2D gels and microarray images to microscopy images and tissue and cell samples. Currently, most of these images are stored on local desktops and CD-ROMs, rendering them not only inaccessible to other researchers, but impossible to analyze and mine in the same manner as alphanumeric data.
Hoping to carve out a new niche market in a field it has dubbed image informatics, Scimagix is launching the Scimagix Application Suite, a set of specialized applications for high throughput analysis and mining of image data. The suite was designed for use with the companys previously released Oracle and Web-based SIMS (Scientific Image Management Solution) database system for the storage and management of image data.
The first module in the Application Suite, ProteinMine, analyzes, searches, and mines 2D gel images. Further modules currently under development are CellMine, an application to analyze, quantitate, search, and mine images of microscopic cellular activity, and TissueMine, an application to analyze, quantitate, search, and mine images of tissue from various sources.
Bob Dunkle, CEO of Scimagix, said that the time is right for his companys solution. Even just three to four years ago, disk space was prohibitive to store large amounts of image data. Now that storage capacity has increased dramatically, its practical, he said.
Indeed, the storage requirements of image data can be substantial. A single 2D gel image can require 1-10 MB, adding up to tens of gigabytes per week for an average proteomics facility. But now that disk storage is cheap, network storage of image data is a real possibility for many companies, albeit one they may not have considered previously.
Theres an element of missionary work we need to do here, Dunkle said. Labs that currently store their images in three-ring binders and CD-ROMs need to be convinced to change their ways, but once they see the potential advantages of the system they are quick to give it a try, he added.
Platform Built on Open Architecture
SIMS is based on an open architecture that supports integration with instruments, databases, and other applications. The analysis and application modules can be combined with other third-party analysis tools to sit on top of the core infrastructure as an application tier. Scimagix hopes to partner with application providers as well as image content providers in the future in order to increase the capabilities of the system.
Image data can be combined with non-image data, although Dunkle said a parser is required to retrieve data from flat files.
The technology behind ProteinMine and the other application modules quantifies and characterizes the color, size, shape, texture, and location of regions of interest in images of different types. Using ProteinMine, for example, proteomics researchers can search a collection of 2D gel images to retrieve gels that have a similar protein expression pattern or a specific therapeutic response. The image-based approach is complementary to statistical 2D gel analysis approaches like Compugens Z3, Phoretixs products, and ProGenesis from Nonlinear Dynamics, Dunkle said.
Scimagix developed ProteinMine in a collaboration with Pfizer Global R&D. Ruth Van Bogelen, who manages the genomics and proteomics group at Pfizer, Ann Arbor, compared it to doing a Blast search on a 2D gel image instead of on a sequence string.
She said the system improves upon existing tools that convert images to numerical data, which can often result in translation errors. People are much more confident if they can look at the original image, Van Bogelen said. She suggested that proteomics groups such as the Human Proteome Organization might be interested in adopting the technology as a standard for storing 2D gel data sets.
The Pfizer group is midway through a year-long process of populating the SIMS database with legacy image data. Were just beginning to see how it will really save us time, Van Bogelen said. While the group is starting out with 2D gels, it plans to eventually add other types of images to the system. As we get our database all set, then were going to add all kinds of other images related to same experiment or compound, so it might be histology or NMR spectra or mass spec spectra.
The Scimagix technology will eventually be incorporated in a larger LIMS system that the group is developing on its own, Van Bogelen said.
In addition to Pfizer, Eli Lilly and four other pharmaceutical companies have adopted the Scimagix system. The company sees the image informatics market reaching $50 million over the next three to five years, growing to $200 million in five to seven years as it reaches the biotech, clinical trials, and medical sectors, and eventually reaching over $1 billion if adopted by the non-science sector.