CRISPR Patent Dispute: applying advanced patent analytics

CRISPR, which is short for Clustered Regularly Interspaced Short Palindromic Repeats, are segments of DNA containing repetitive base sequences. In recent times a technology known as CRISPR/Cas9 has been developed to allow the addition or removal of genes, so enabling the editing of the genome of a cell.

This ability to edit a genome is a big deal, and not surprisingly was the choice of Science magazine for the Breakthrough of the Year in 2015.  

Not surprisingly with such a potentially important invention, there have been disputes over ownership. In particular there has been a high profile dispute between theBroad Institute (MIT) and University of California.  The University of California had filed first, with a priority of date of May 2012 for their published applicationUS20140068797, which is yet to be granted.

 The Broad Institute first filed a little later in December 2012, but then pushed the examination process so that the first of their patents was granted in April 2014. Since then the Broad Institute has gone onto to receive more than 50 granted patents for this technology.

In response, the University of California launched a patent interference process to determine who was the first inventor of this technology.  

On 15th February the USPTO handed down its decision on this case, finding that there was no inteference-in-fact between the patents, namely US20140068797 and a group of patents filed by the Broad Institute in December 2012:

..we conclude that the parties’ claims are not drawn to the same patentable subject matter and that there is no interference-in-fact between them.

There is a 51 page judgment to this effect, and here at Ambercite we will leave the judgement and others to explain the reason for this.

But we were curious - what would advanced patent analytics tell us about this case? 

To answer, we will use our the Family Cluster Searching patent analytics tool. This can find similar patents to one or more 'seed patents'. Essentiallly all of the patent citations in the area of a patent or a group of patents are combined together to identify these similar patents. Because every patent in this area may have its own search report, this network represents the opinion of every applicant and every examiner in this area about which individual patents are similar. 

We call it 'Family Cluster Searching' because each patent in our network is  in fact a family of patents. And so even a single point in the network can combine the opinions of say the PCT patent examiner, the US examiner, the EP, JP and even Chinese examiners. A lot of data is combined in this 'collective wisdom approach, and this helps to produce a statistical robust approach. 

So what can advanced patent analytics tells us about this case?

Was the potential for litigation predictable?

Family Cluster Searching comes with the ability to identify similar patents to one or more starting patents. To answer this question, we took the Broad Instittute patents mentioned in the litigation, and ran a search for the most similar patents.

This query looked like this:

Note that we had changed the application # 14,704,551 to its published version US2015024710 - and that we had requested the most similar 2000 patents

And what did we find?

Not surprisingly, many of  most similar patents found were are filed by the Broad Institute, such as five out of the top 10 shown below.

And this list goes all the down to position #1993

But the real question is, as hinted by the image above, is 'does this list include the University of California patent?

To answer this question, we applied a filter to look for patents filed by the University of California:

Which produces this list of patents:

In fact the full list is somewhat longer than this, but at top of the list is  WO2013176772, Methods and compositions for rna-directed target dna modification and for rna-directed modulation of transcription .

This is in the same family as US20140068797, which is the  published form of the litigated patent 13/842,859. And according to this analysis, there is a total similarity score of 370 to the Broad Institute patents shown. This is a very high figure, and suggest strong similarity between the Broad Institute patents and the University of California patent. 

So no surprises about the litigation then - they are in very similar fields. The University of California patents was the 3rd most similar patent

Of course, this is a bit of a 'so-what' analysis - a number of different pathways would have produced a similar outcome, including the litigation history itself. But - this analysis was quick and simple, and so a useful example of the potential of patent analysis to at least identify the potential for litigation (or licensing), based on patent numbers alone. This can reduce the time and cost needed for other, more time-consuming and expensive forms of analysis

Who else filed the most similar non-Broad Institute patents?

To answer this question, we can filter out the Broad patents from the results using the following filter

Which leads to the following list of similar patents:

So a big range of applicants. From these results as a whole, we can make a range of analyses - such as the analysis immediately below.

Who has the most commercially similar patent portfolio?

We can predict this from our Licensing Potential metric, again which is based on the Broad Institute patents listed in the litigation:

Sangemo Therapeutics describes itself as "Developing the most advanced and precise genomic medicines", while Arbustus Biopharma is primarily focused on Hepatitus B, that is also  developing a pipeline of Non-HBV Assets that leverage our expertise in RNA interference (RNAi) therapeutics

Perhaps what is most surprising is that the the patents belonging to the University of California have a comparably low Licensing Potential compare to some other companies.

We should note that this analysis, like any analysis, is limited by the assumptions used. In this case one of the key assumptions is that the Broad Institute patents were only those listed in the litigation. These were all the early patents filed by the Broad Institute, and they have filed many since. Ideally we would rerun this analysis with the more recent Broad Institute patents, but such an analysis would sit outside the scope of this blog.

What are the most similar prior art patents? 

By looking for earlier priority date patents filed prior to December 2012, and say looking for patents filed after 2002 to allow for the rapid advances in this field, we would suggest that the most similar prior art patents would be:

Patent family (representative patent)	Rank	Similarity	Owner	Title	Priority date	Citation
WO2013176772	15	102.5	UNIV CALIFORNIA	Methods and compositions for RNA-directed target DNA modification and for RNA-directed modulation of transcription	2012-05-25	Known
US7045676	21	72	GENZYME CORP	Transgenic animals secreting proteins into milk	1986-04-09	Known
US4873316	23	66	BIOGEN INC	solation of exogenous recombinant proteins from the milk of transgenic mammals	1987-06-23	Known
US4897355	24	65.5	SYNTEX INC	N[ omega ,( omega -1)-dialkyloxy]- and N-[ omega ,( omega -1)-dialkenyloxy]-alk-1-yl-N,N,N-tetrasubstituted ammonium lipids and uses therefor	1985-01-07	Known
WO2013142578	25	63	VILNIUS UNIVERSITY	RNA-directed dna cleavage by the Cas9-crRNA complex	2012-03-20	Known
US9005973	26	62.5	SANGAMO BIOSCIENCES INC	Targeted genomic modification with partially single-stranded donor molecules	2010-02-09	Unknown
WO2014089290	27	62	SIGMA ALDRICH CO LLC	CRISPr-based genome modification and regulation	2012-12-06	Known
WO2014011901	28	60.5	SANGAMO BIOSCIENCES INC	Methods and compositions for delivery of biologics	2012-07-11	Unknown
US8889394	29	59.5	EMPIRE TECHNOLOGY DEV LLC	Multiple domain proteins	2009-09-07	Known
US7888121	31	57	SANGAMO BIOSCIENCES INC	Methods and compositions for targeted cleavage and recombination	2003-08-08	Unknown

Not surprisingly, the litigated University of California patent was at the top of the list, but there were many others - and this is only a top 10 list. 

Note too that the list comprised a mixture of 'Known' and 'Unknown' citations. An 'Unknown' citation is a patent family which has never been cited against the patent family being examined. While not all unknown citations in a Family Cluster search are relevant, many are. 

Next steps?

This is what we could learn from just Family Cluster Search. Just like any other searching system, we recommend running a range of different queries to end up a range of different results - you may be surprised what you can find.