Towards a graph genome
Yesterday at Festival of Genomics, London, Seven Bridges CEO Deniz Kural joined a panel discussing implementations of graph genomes, alongside Erik Garrison from the Wellcome Trust Sanger Institute and Andy Yates from EMBL-EBI.
— Adrian Alexa (@_aalexa_) January 31, 2017
We’ve written up a few take-home points from the ensuing discussion:
1. Not everyone is familiar with genome graphs (yet)
The initial questions from the audience asked the panelists to describe and visualize a graph genome. Unfortunately owing to the session format the panel wasn’t able to put up a slide. But we have a brief overview here, as well as some more detailed explanations on building a graph reference and aligning reads to the graph.
We’ve also made a short video to illustrate how aligning sequencer reads to a genome graph can help to improve genomic variant calling.
2. We’re moving towards graph genomes, but…
…as ever, work remains to be done.
First, we can do more to ensure adoption of graph genomes. Andy Yates noted that while a linear reference has limitations, it is familiar, and conceptually easy to think about. For this reason, we intentionally build our graph genome tools to be interoperable with standard file types and existing workflows.
Second, implementing a graph genome is only half the battle. To make the implementation practical requires a robust means of annotating with phenotypic and clinical data. We are exploring ways to do this in our research collaboration with the US Department of Veterans Affairs, as part of the ongoing Million Veteran Program.
Finally, we’ll need to clearly demonstrate utility of graph genomes. Deniz highlighted some of our ongoing work in this area, including the projects our scientists are working on with leading research institutes, and our commitment to benchmarking performance of our graph genome tools, in terms of accuracy, runtime and cost.
— SevenBridgesGenomics (@SBGenomics) January 31, 2017