Qiagen this week said that its recent acquisition of SABiosciences brought more than just a line of PCR assay panels, but also contributed bioinformatics know-how and expertise that will help it advance its biomarker discovery efforts.
In a conference call to discuss Qiagen's second-quarter earnings this week, CEO Peer Schatz said that the bioinformatics capabilities the company obtained from the acquisition — particularly when coupled with its DxS companion diagnostic division — have been “a much stronger asset than expected.”
Qiagen acquired SABiosciences late last year for $90 million in a deal that gave it a line of around 100 disease- and pathway-focused real-time PCR-based array panels. That purchase followed the acquisition three months earlier of DxS for $95 million.
Xiao Zeng, who was formerly vice president of research and development at SABiosciences and now holds the same title at Qiagen, told BioInform this week that the “strategic reason” that Qiagen purchased both companies was to meet needs at both ends of “the pharmaceutical drug discovery and development pipelines”
While the DxS diagnostic products are focused on the different phases of clinical trials, SABio’s products focus on early drug development, which includes discovery, final candidate selection, and drug toxicity and efficacy studies.
In order to screen for biomarkers to include in its RT2 Profiler PCR Array test panels, SABio developed a text-mining tool that Zeng called the “foundation of everything SABio did and is still doing.”
The tool uses natural language processing techniques to extract and process information from the scientific literature in order to identify potential markers. The extracted information is stored in a database.
In developing the tool, Zeng said that SABio was trying to solve the longstanding problem in bioinformatics of integrating old and new genetic information.
“Before microarrays even existed, people [were] doing biological research for decades,” he said. “That information and knowledge has accumulated over the years [in the literature], but how do you reconcile the new information with the old knowledge?”
He added that over the years, SABio's developers have improved the “dictionary and semantic rules” of their NLP tools in order to limit false positives.
“For the entire knowledgebase, we would incorporate all the available public domain databases with our data-mining tools and some licensed ones, such as the Human Protein Reference Database from Johns Hopkins University,” he said.
Zeng said that SABio offers pathway- and disease-specific applications so that researchers don’t have to canvas the entire genome to study a single disease or pathway.
“If you know that you are studying cancer and you are most interested in apoptosis and how that affects your drug or disease treatment…we basically say, this is the package of genes that you need to concentrate on,” he said.
Once the researchers have the data they need, Zeng said that SABio provides web-based bioinformatics tools to sort through it. The company also offers free software that draws on the SABio knowledgebase to provide information about interactions between multiple genes and allows users to analyze relationships in gene pathways and processes.
“We have the usual classification algorithms everyone else is using,” he said. “We have also developed support vector machines and Bayesian latent factor based models, which are more computationally expensive to run and not very common for regular microarray users.”
SABio currently has nine employees in its bioinformatics group — up from six employees prior to the acquisition. Zeng said that the company plans to hire additional bioinformatics staff.
[ pagebreak ]
The Hunt for Biomarkers
When Qiagen acquired SABio, it noted that it expected its products to be complementary to DxS, which is developing companion diagnostics such as its Therascreen KRAS mutation kit.
So far, DxS has signed an agreement with Bristol-Myers Squibb and its partner ImClone Systems to develop a KRAS test kit for their co-marketed colorectal cancer drug Erbitux, and has signed a similar deal with Amgen for its colorectal cancer drug Vectibix.
The TheraScreen assay detects KRAS mutations in codons 12 and 13. Last year, the US Food and Drug Administration updated the labels of the two drugs to recommend that they should not be administered to colorectal cancer patients with these mutations since the drugs will not benefit them.
The test is currently being reviewed by the FDA.
Zeng noted that SABio's tools could help Qiagen's DxS group develop additional tests using a similar model. He said the company's bioinformatics group has already identified a collection of additional biomarkers that could help companies develop targeted drugs or help tailor therapies to specific patients.
Zeng said that these biomarkers would be most useful for researchers studying specific diseases like cancer, for drug efficacy and targeted drug studies in the pharmaceutical industry, and in clinics to help physicians make treatment decisions.
According to Zeng, Qiagen is planning to continue the development of bioinformatics tools to make the process of discovering biomarkers more efficient to identify, for example, differences between a normal sample and a cancer sample.
“There are a lot of mathematical and statistical models that you can build to look at those differences … a lot of biomarkers have been there in the literature, but unfortunately very few have been followed through all the way to clinical use,” he said. “The final goal is still to make a clinically or pharmaceutically useful biomarker. In order to get that down, you must have a better algorithm to look for those biomarkers.”
He also said Qiagen plans to develop a tool that will convert “heat map” information from microarrays into a useful score that will eventually help clinicians make better treatment decisions. The same score could also be used for drug screening, or for monitoring drug engagement of targeted pathways, he said.
Zeng said that the company is using its biomarker discovery and validation tools to develop new PCR arrays that will be similar to its RT2 Profiler PCR Array but will contain different “information content.”
“In our current version, we only provide a list of assays that are relevant for a given pathway or disease, but we do not tell the customer how to interpret the profile,” he said “In our next version … we will include a software package to convert the expression profile into a score.”
For example, he said that the company is releasing induced pluripotent stem cell PCR arrays in September. Researchers can use the array and its companion software to get “a likely score based on the expression files of some carefully selected and validated key genes” to determine whether they are working with iPSCs or not.