Researchers at MIT have compared nucleotide sequences in US patents against the human genome to find that nearly 20 percent of human protein-encoding sequences are protected by patents.
"We were actually surprised at how high the number was," said Fiona Murray, a co-author of the study, which was published in the Oct. 14 issue of Science. "Although there has been a lot of discussion of gene patenting it often focuses on particularly well-known examples such as the BRCA1 gene. We were surprised to find such extensive patent coverage."
In an e-mail interview with BioInform, Murray said that the study was the first step in a "much more thorough empirical analysis of the impact of gene patents on scientific and commercial progress."
Murray, an assistant professor at MIT's Sloan School of Management, and Kyle Jensen, a graduate research assistant in MIT's Bioinformatics and Metabolic Engineering Laboratory, used Blast to compare patented sequences to NCBI's RefSeq and Gene databases. They found that 4,382 of the 23,688 genes in NCBI's gene database were claimed in 4,270 patents that are owned by 1,156 assignees. This was a conservative estimate, Murray stressed, because they only looked at patents that explicitly claim gene sequences, and excluded patents claiming amino acid sequences.
Incyte alone owns patents covering 2,000 genes, "mainly for use as probes on DNA microarrays," the authors noted. Other top gene-patent holders include the University of California, Isis Pharmaceuticals, SmithKline Beecham (now GlaxoSmithKline), and Human Genome Sciences.
It is unclear what if any impact this level of patenting has had on research. As the authors note, critics of gene patents have argued that overly broad patents "might block follow-on research," or may cause "an anticommons effect, imposing high costs on future innovators and underuse of genomic resources." These arguments have been difficult to prove or disprove, however, without concrete data on the extent of gene patenting.
"Much of the current policy is based on anecdotal evidence and incomplete data," Murray told BioInform. "It has traditionally been very difficult to really build up a full picture of patenting of human genes, but we were able to combine bioinformatics methods more typically used in scientific research with various patent databases and this has allowed us to build up this data."
Murray said that she now plans to examine some of the "key issues" in the debate over gene patenting, including "who owns the patented genes, how do they enforce their ownership rights, is there a lot of costly litigation or do the various parties come together and establish efficient licensing agreements rather like the semiconductor industry."
One "immediate impact," Murray noted, is that this study should "raise questions about the ways in which traditional genetic information (in databases such as Genbank) is connected to patenting information." More explicit linking between these resources "will encourage transparency and reduce the costs to future researchers of complex and costly patent searches," she said.
Another interesting finding, Murray said, is that among genes with many patents, "there are often patents on isoforms that are extremely similar and that at least appear to effectively claim the same gene." This, she said, "raises a number of questions about whether the examination process is effective and how we should search for and define the prior art."
Bernadette Toner ([email protected])