Completed. Last week, we completed the recruitment for one of the more visionary projects at CHOP that we are involved in – the epilepsy pilot project of the Genomics Research and Innovation Network (GRIN), a collaboration between CHOP, Boston Children’s Hospital, and Cincinnati Children’s Hospital. Here is my personal take on GRIN and why I think that forming, building, and expanding GRIN is so important in our current research environment.
A brief disclaimer. As usual when I am writing about major initiatives, I would like to emphasize that this blog post expresses my personal views, not the official views of the Children’s Hospital of Philadelphia or any of the other partners involved. To me, it is important to point this out as our emerging GRIN network already has many different facets. It is impossible for me to give equal credit to each aspect of GRIN and I’m limiting myself to the epilepsy pilot project that we are involved with.
The GRIN foundation. In 2015, the CEO’s of CHOP, BCH (Boston), and CCHMC (Cincinnati) came together and agreed to fund a pilot project to jointly create a sustainable research infrastructure for pediatric genomic research, a collaborative project that explores the possibilities of building and maintaining a joint network for pediatric genomic research. GRIN is a self-funded, self-sufficient initiative of the three hospitals, a strategic investment into the future of applying genomics to pediatric healthcare.
Making GRIN possible. There are at least two things that are surprising about the decision to form such an initiative. First, it is somewhat unusual that such an initiative is proposed at all – there are more than enough reasons for these three medical centers to not collaborate. Bluntly speaking, we are competitors. We compete for federal research funds, our ranking in the U.S News & World Report, the best staff, and most advanced pediatric specialty programs. Secondly, a program like GRIN may seem superfluous – all three centers are already involved in cutting edge genomic research and there are a number of historical research connections between the three sites. So why would we create a network like GRIN?
Explaining GRIN. Understanding GRIN requires a shift in perspective. It requires us to take a step back and examine our current shortcomings and bottlenecks in genomic research. I usually point out the following two issues. First, we are sitting on data silos. With the exception of a few large centers, today’s genomic data in a clinical and research context is fragmented, split apart by a myriad of firewalls and regulatory frameworks. Even though we like to think about novel ways to analyze large datasets, we rarely ever have the opportunity unless you belong to a small, privileged class of researchers. Even though we frequently invoke our intention to engage in data sharing, there is not really an incentive for this. Secondly, genetic data is simply one data type that is important to us. If we consider our mission to understand disease in a more comprehensive context, our analysis of phenotypic data is stuck in the pre-industrial age. While we are able to provide sophisticated analysis tools for genomes, our ability to integrate clinical data barely makes it past a two-by-two Excel table. GRIN is about creating an opportunity to overcome these issues and to provide a template that we can expand to other centers in the future. GRIN is about taking all the necessary steps to build a rich and sustainable sharing culture between the participating centers. GRIN is about creating a novel opportunity space to ask questions that we don’t even know yet.
The epilepsy pilot. In the overall GRIN architecture, we are conducting one of the three pilot projects. Together with Eric Marsh and Holly Dubbs (CHOP), Ann Poduri and Beth Rosen-Sheidley (BCH) and Barbara Hallinan and Tracy Glauser (CCHMC), we are working on a small, but unusual pilot for trio exome sequencing in unsolved epileptic encephalopathies. A project that should be a scalable template for gene discovery and data sharing in rare diseases for the years to come. By completing the recruitment of our patient-parent trios, we have now finished the more conventional part of this project. What follows then is the truly novel aspect of GRIN. Once exome sequencing is completed, the data will be released into a joint cloud environment. The data does not belong to any single center, it is joint GRIN data. This will be combined with extracted phenotype data from electronic medical records (EMR) of our three centers, giving a serious attempt at the “EMR extraction problem” that we are currently facing – we are sitting on terabytes of electronic patient data and do not understand it. Again, GRIN is about the opportunity to mine this data in combination with collaborative gene hunting.
Five things. Let me sum up my fascination of GRIN in five things that I think make a true difference in the way that this collaborative may impact on how we perform research in the future.
1 – Sustainability. A major aspect of GRIN is providing a model for demonstrating sustainability, being fully aware that traditional research funding can only ever accomplish parts of what we want to achieve and may even be counterproductive by throwing us back into silo mode. In this respect, GRIN has more in common with a start-up company than a traditional research project. GRIN has an entire section of its project dedicated to interacting with industry and exploring how collaborative genomic research can become so imbedded in the architecture of our hospitals that it becomes self-perpetuating. Think of GRIN as the new data platform that is worth investing in – the ability to have the joint data of the three prominent US pediatric centers at your fingertips with all the regulatory issues already taken care of.
2 – Culture. GRIN has already accomplished something that may not be that obvious at first glance. Three large institutions have large different cultures when it comes to our approaches to regulatory issues, data ownership, tools for data storage and analysis, and personalities (yes, hospitals have personalities). In an ideal world, all of this should not matter and GRIN should proceed in a streamlined top-to-bottom approach. In reality, however, these differences matter significantly and they always have in collaborative research. What we have accomplished in our first year is to understand these differences and move past them. While some historical and cultural differences remain, we have put structures in place to perpetuate the collaborative aspect of GRIN.
3 – Sharing. Again, GRIN is a shift in thinking. It is not about what this network can do for each one of us individually, but about how our research may benefit the network. Our entire epilepsy genetics research at CHOP that extends well beyond our pilot project is built on the GRIN framework, it is “powered by GRIN”. Every epilepsy patient we consent and family we recruit will be part of GRIN. On purpose, we tied ourselves so close to GRIN that it will be difficult to “brexit”. This way, whatever we do, data sharing will be part of our DNA and the data from any future epilepsy genetics project will stream into GRIN, creating novel opportunities for discovery.
4 – Possibility. I would like to think of GRIN as the possibility for a gigantic accelerator. One major problem we face in rare disease research is the inability to truly connect data. Yes, we can share Excel sheets of our favorite genes and phenotypes, but do we really have the possibility to comprehensively mine raw data across sites and apply novel tools to squeeze as much information out of data as possible? We rarely get this opportunity, which is why the GRIN cloud sharing approach is so important. We sometimes joke that everybody will know in three years what an S3 bucket is (hint: it’s the virtual place in the Amazon cloud where a cloud-based exome is located). Our genomic dataset is becoming too enormous and only distributed data storage solutions will be able to hold them. In GRIN, we are already exploring how we can use these resources for collaborative research and not only storage bins.
5 – Pediatrics. Finally, GRIN is all about pediatrics. While I may be stating the obvious, I would like to emphasize how important it is to put pediatric research at the forefront given how far we are lagging behind. Pediatrics will always be slow and intense given the complexity of our patient population with rare diseases and multiple comorbidities. GRIN is one of the many initiatives that tries to turn this disadvantage into a major asset, helping us to find better diagnostic strategies and treatments. The same ideas that drive the GRIN collaborations currently fuel various commercial and academic initiatives in the rare disease community – we may finally be at a point where our ability to handle data helps us turn the many weaknesses of pediatrics into our ultimate strength.