This week’s Video Tip of the Week follows on last week’s chatter about the Internet of DNA. As I mentioned then, the Beacon tool we touched on was going to get more coverage. So this week’s video is provided by the Beacon team, part of the larger Global Alliance for Genomics and Health project (GA4GH).
The talk was called “Stable Reference Structures for Human Genome Analysis” and it was important for me to see this. I’ve been wrestling with some of the literature (linked below) that describes ways to represent genome variations among massive numbers of humans. It really helped me to hear it described and shown as cartoons on slides that were less like equations. And how this will play out in graphs and visualizations with software tools is of particular interest to me.
So one branch of the Data Working Group of the GA4GH is tasked with how to represent the variations as multiple paths as graphs, instead of the one linear reference genome we think of today. It has to accommodate many types of variations–inversions, deletions, duplications, as well as just SNPs. So, as the kids say today, it’s complicated. But we have to figure it out. Stay tuned, I’m sure we’ll be talking more about this in the years to come.
Another branch of this project is tasked with trying to figure out how to share genomic data among all the international producers of this data. If we can’t share the data, we won’t be able to look at the variations among humans and learn from them, nevermind display them. This has additional layers of social and legal complexity we are just beginning to face. As a first pass at sharing this data, a “Beacon” system has been implemented to help researchers locate variations of interest to them.
You should read up on the whole Beacon philosophy and see its current implementation at their site. From what I gather, it is a minimal way to share genome information, without incurring privacy and consent barriers that might be hit if you were pulling down a whole genome. You can query any site that implements a Beacon to ask: do you have a variation at this position? And the Beacons can respond with “yes” or “no”. If there are useful variations, you can then pursue them from there, and if you need access to more you can go through the channels then. But at least you’ve possibly found some needles in some haystacks that you might not have known about otherwise.
The Beacon team has done a short video explaining this. It has no audio, just explanatory text with the graphics. Marc Fiume gave me permission to embed it here.
The “Beacon of Beacons” aggregates the query to send it out to all the known Beacons. You can use it today to search for this kind of data. The video also notes that you can cloak the name of the institution to protect patient privacy.
I have been more acutely concerned about genomic privacy issues than some of my cohorts in this arena. And I fully accept that there will not be privacy–what I want is protection from misuse of the information, which I find lacking in the US legal framework right now. That said, I think that Beacon is a nice work-around for that. If I had a variant of concern, I could ping these other sites to see if others had it. Or vice-versa. But the framework under which the donor of that material provided the data would not be pierced. This makes total sense to me, and I can accept this strategy.
Sharing the genomic data from sequenced individuals is going to be tricky and complex. But I’m keen to see the GA4GH group tackle it. I like several of the directions that I’ve seen so far. But right now–check out Beacon. Implement one if you have this kind of data, and let’s see if it works.
Global Alliance for Genomics and Health: http://genomicsandhealth.org/
Beacon (project details page): http://ga4gh.org/#/beacon
Beacon of Beacons (where you would do a search): http://ga4gh.org/#/beacon/bob
Nguyen N., Glenn Hickey, Daniel R. Zerbino, Brian Raney, Dent Earl, Joel Armstrong, W. James Kent, David Haussler & Benedict Paten (2015). Building a Pan-Genome Reference for a Population, Journal of Computational Biology, 150107093755006. DOI: http://dx.doi.org/10.1089/cmb.2014.0146
During David Haussler’s talk, he also referenced these papers:
— OpenHelix Staff (@OpenHelix) February 18, 2015