Making good on a promise to "aggressively protect" its patent portfolio, biomarker-discovery firm Health Discovery Corp. this week filed patent-infringement lawsuits against Ciphergen Biosystems and Equbits in the United States District Court in Marshall, Texas.
The suits, which accuse the firms of selling software that infringes on HDC's support vector machine technology, are a relative rarity in the field of bioinformatics, which has historically been far less litigious than other biotech sectors.
While the broader impact of the lawsuits on the field of bioinformatics is unclear, it could affect many groups using SVM technology for life science research. SVMs, machine-learning algorithms that are effective in classifying large data sets, have proven to be valuable tools in bioinformatics and biomedical research: A PubMed search on "support vector machine" or "support vector machines" generates 836 hits dating back to 1998. A similar search in the journal Bioinformatics brings up 307 papers.
James Barta, an associate with law firm Senniger Powers, noted in an e-mail to BioInform that "while the term 'bioinformatics' receives scant mention in the patents asserted by HDC, it is clear that these patents and the lawsuits have relevance to organizations and individuals in the field of bioinformatics."
Bioinformatics groups using SVMs and other forms of machine learning "may be concerned about becoming a defendant in a patent infringement lawsuit, and rightly so," Barta added. In addition to HDC's IP portfolio, Barta said that a search of the US Patent and Trademark Office database identified nearly 200 patents that refer to SVMs issued to major IT players like IBM, Hewlett-Packard, and Mitsubishi Electric Research Labs.
"While the term 'bioinformatics' receives scant mention in the patents asserted by HDC, it is clear that these patents and the lawsuits have relevance to organizations and individuals in the field of bioinformatics."
In two separate suits filed on June 26, HDC alleged that Ciphergen's ProteinChip software, Biomarker Patterns software, and CiphergenExpress software and Equbit's Foresight software infringe three of its SVM patents: US Patents 6,427,141; 6,760,715; and 6,789,069 (see box for details of the patents named in the complaints).
The suits accuse both Ciphergen and Equbits of selling infringing products "to analyze protein expression data for discovery of biomarkers for use in diagnosis and treatment of disease, including various types of cancer."
The complaints ask that both Ciphergen and Equbits "be enjoined from making, using, offering for sale, or selling any product that infringes one or more claims of the patents-in-suit (and inducing or contributing to any other person doing the same)," along with compensatory damages, enhanced damages, attorney's fees, and other costs.
Neither Ciphergen nor Equbits returned calls seeking comment on the suits.
Barta said that the claims of the '141 patent "do not explicitly recite the analysis of biological data," and could be enforced against products in other disciplines, but said that the claims of the other two patents refer specifically to biological applications.
Furthermore, he noted, the '715 patent includes among its claims "a method of diagnosing disease" using SVMs, and "a diagnostic device comprising genetic probes that specifically hybridize to genes identified as being associated with a disease by multiple learning machines."
Patents Identified in the Lawsuits
US Patent 6,427,141. Enhancing knowledge discovery using multiple support vector machines. Inventors: Stephen Barnhill. Assignee: BIOwulf Technologies.
Protects a system and method for "enhancing knowledge discovery from data using multiple learning machines in general and multiple support vector machines in particular," according to the patent abstract. Multiple support vector machines, "each comprising distinct kernels, are trained with the pre-processed training data and are tested with test data that is pre-processed in the same manner."
US Patent 6,760,715. Enhancing biological knowledge discovery using multiple support vector machines. Inventors: Stephen Barnhill, Isabelle Guyon, Jason Weston. Assignee: Barnhill Technologies.
Covers a method in which multiple support vector machines are used to "extract useful information from vast quantities of biological data," according to the patent abstract. The method includes pre-processing of training data and test data. The training data is used to train the learning machine, after which the success of the training is tested using the test data. After the training has been confirmed, "live biological data can be pre-processed then input into the identified support vector machine that provides the optimal solution for extraction of useful information."
US Patent 6,789,069. Method for enhancing knowledge discovered from biological data using a learning machine. Inventors: Stephen Barnhill, Isabelle Guyon, Jason Weston. Assignee: BIOwulf Technologies.
Protects a method in which a "learning machine" is used to extract information "from vast quantities of biological data," according to the patent abstract. In the preferred embodiment, the learning machine is one or more support vector machines.
"These claims target both the user of a software product that embodies the SVM technology defined in these claims and the manufacturer or developer of a device that embodies the SVM technology defined in these claims," Barta said.
Therefore, users of third-party SVM software, as well as those developing SVM software, could be infringing HDC's patents.
For those considering the purchase of an SVM product for analyzing data, "insist on an indemnification clause in the purchase or license agreement," Barta said. This clause should hold the vendor financially responsible for certain types of damages, claims, or losses arising from the use of the product, including patent infringement actions.
Barta said that developers of SVM-based bioinformatics tools should consult an IP attorney "regarding the risk of patent infringement for the HDC patents as well as any others."
While acknowledging that "there has been a paucity of patent litigation in the bioinformatics field," Barta noted that that the developers and users alike should be prepared. "Ignorance is not a defense to infringement," he said.
It is unclear whether academic researchers face the same risk as commercial developers. William Stafford Noble, an assistant professor of genome sciences at the University of Washington who specializes in machine-learning bioinformatics methods and has collaborated with Jason Weston, an inventor on two of the patents named in the suit, said that even though SVMs are very common in the field, "I don't really have a sense of how [the HDC patents] would impact academic research. There are so many SVM implementations that are free on the web that it seems like they would have a hard time trying to prevent those from getting distributed."
Noble said that he doesn't plan to license the SVM technology from HDC. "We're just writing papers. I'm not making any money off of it," he said.
Building — and Enforcing — the Patent Portfolio
Health Discovery Corp. got its start in 2003, when telecommunications firm Direct Wireless, a publicly traded firm on the over-the-counter bulletin board, acquired bioinformatics consultancy the Barnhill Group and software firm Fractal Genomics with the goal of building a biomarker-discovery company [BioInform 09-08-03].
Founder Stephen Barnhill held several patents covering neural networks and SVMs for life science research, and the company added to that in 2004 by acquiring the rights to a portfolio of 71 issued and pending patents on SVMs and other machine-learning tools with applications in diagnostics and drug discovery. These included several patents naming Vladimir Vapnik and Isabelle Guyon — widely considered to be pioneers in the field of SVMs — as inventors (a complete list of the company's issued and pending patents is available at http://www.healthdiscoverycorp.com/intellectual.html).
As of March 31, the company valued the patent portfolio at $3.5 million, according to filings with the US Securities and Exchange Commission.
Last year, the company said that it had expanded its rights to that portfolio and that it had launched "a new licensing and product development program" around the patented technology [BioInform 5-23-05].
The new program included "an aggressive campaign to pursue licensing arrangements with corporations and institutions that are already using [Health Discovery's] patented technology," the company said in a statement at the time, adding that it "intends to aggressively protect the rights granted by this significant and mature patent portfolio."
Last week's suits are the first that the company has filed, but Barnhill said in a statement this week that they are "one step in HDC's efforts to protect its valuable intellectual property." He added that the company plans to continue its internal biomarker discovery initiatives, but "will actively pursue any company that does not respect our IP rights."
Daniel Furth, principal financial officer of HDC, told BioInform that the company's IP strategy "has been license-oriented and primarily in the biotech space, and that's going to continue to be our licensing strategy."
HDC recorded $105,000 in licensing revenue in 2005 and its cash holdings were $522,910 as of March 31.
The company said this week that William Quirk, a member of its board of directors and its largest individual investor, intends to purchase up to $100,000 worth of HDC stock "in support of HDC's strategic direction."
— Bernadette Toner ([email protected])