Clinical Data's Icoria subsidiary said this week that it is partnering with IO Informatics for the final leg of a five-year, $11.7 million grant from the National Institutes of Standards and Technology's Advanced Technology Program.
The ATP grant — the largest ever in bioinformatics — was originally awarded in 2002 to Paradigm Genetics and Lion Bioscience to develop the Target Assessment Technologies Suite, an informatics framework that was expected to "improve data integration, transform data from different platforms into one coherent data set, and analyze these data sets to identify gene function or mechanism of action — and potential product research targets — rapidly, reliably, and efficiently," according to the NIST project brief [BioInform 06-17-02].
A lot has changed since then, however. In August 2004, Paradigm switched its focus from genomics to systems biology and renamed itself Icoria, and was acquired by Clinical Data last December. Meanwhile, the data-integration component of the project changed hands several times: In October 2004, Lion withdrew from the project and was replaced by Agilent Technologies [BioInform 10-11-04]. Now, Agilent has also withdrawn from the project and IO Informatics has stepped in.
Nevertheless, the project has remained "remarkably intact in terms of its objectives, despite all the changes around the partnership on the grant," said Tom Colatsky, chief scientific officer at Icoria. Colatsky noted that the "initial phase" of the grant, which is just coming to a close, was focused on generating "very complex, high-dimensional, high-quality data sets" for gene-expression, metabolomics, and histopathology experiments, and was primarily Icoria's responsibility.
"We didn't even exist at the time this grant was initially awarded."
In that sense, he said, "the timing is very, very good" for IO Informatics' involvement, because the second two years of the project will focus on data integration and software development.
Colatsky added that the project has also emerged unscathed from Icoria's evolution over the last three years. "I think everything that we have been doing to this point — the integration of multiple -omic data streams, the reduction of high-dimensional data sets — all of that is extremely relevant, maybe even more relevant, given that Clinical Data now is our parent company," he said. He noted that Clinical Data's acquisition of Genaissance in October placed new data related to haplotyping and polymorphisms in Icoria's hands, and "all this fits very nicely into the kind of tool structure that we envision for the products coming out of the ATP grant."
Colatsky said that Clinical Data's recently announced plan to split its business into two groups — Clinical Data Molecular, which will comprise Icoria and Genaissance; and Vital Diagnostics, which will consolidate its previously owned clinical subsidiaries — is not expected to have an impact on Icoria's operations.
As for IO Informatics, a startup based in Emeryville, Calif., "We didn't even exist at the time this grant was initially awarded," Patricia Rougeau, the company's CEO, told BioInform. "It's a wonderful validation for us that our approach was compared with everything else on the market, and we were chosen by Clinical Data to partner with them on the grant," she said.
IO's data-integration technology, called Sentient, will now serve as the data-integration backbone of the project, rather than that of its predecessors — Lion's SRS and Agilent's Synapsia Workbench.
Sentient uses an object-based approach, rather than a relational database approach, which enables a "data-centric rather than application-centric" integration platform, Rougeau said. The system relies on metadata that is either supplied by the user or generated automatically, which allows users to "ask questions across that data as if it were all in the same format, even though some of it might be coming from instruments, some of it might be coming from relational databases, some of it might be images."
Sentient's metadata format is based on the same "triple" format as RDF, Rougeau said, which makes it compatible with other systems using semantic web standards.
A US patent covering the Sentient technology, No. 6,988,109, "System, method, software architecture, and business model for an intelligent object based information technology platform," was granted on Jan. 17.
Colatsky called Sentient "an extremely robust platform for integrating disparate data streams," and said that IO has been "a very responsive partner when it comes to thinking about how you can reduce dimensionality, and how you can extract emergent behaviors from situations where if you simply had taken one data set after another, after another, and tried to find something, you would lose some value."
Both partners stressed that their collaboration has just begun, however. Representatives from the two firms began meeting just this week to establish use cases and identify the specific data streams that will be integrated with the Sentient platform. "For example, how many instruments do they want to take data from, how many different data types do they want to take data from? That's part of the use cases that are going on right now," Rougeau said.
The data that has been collected so far is complex. Colatsky said that there are three primary data sets: an analysis of hepatotoxicants using whole-genome expression arrays; metabolomics analysis for around 700 biochemicals; and "quantitative tissue feature measures" for "fairly coarse" histopathology data that includes "14 to 24 individual parameters that are being tracked and annotated."
Colatsky said that Icoria is currently preparing the last of those data sets to deliver to IO Informatics. The next project milestone will be the development of prototype "data coherence" tools that "cross -omic platforms," he said.
Icoria's commercialization path for the technology is still uncertain, but Colatsky said that this is by design. "It's broadly defined," he said, "and I think the reason for that is that the technology is changing so rapidly. We need to be responsive to the markets." He did note that the company never planned on "shrink-wrapping something and selling it," however. "It was a question of applying it in a commercial setting — whether it was going to be applied as part of IO informatics and some of the functionality that we identified and prototyped got built into Sentient, or if we simply used the tools ourselves in discovery or in research collaborations and partnerships — all of that would represent a viable avenue for commercialization of the product," he said.
IO's commercial goals for the project, on the other hand, are much more concrete. "This builds on the product that we already have on the market today," Rougeau said, "particularly in the personalized medicine and pharmacogenomics areas for biomarker discovery."
— Bernadette Toner ([email protected])