Genomic Medicine: Bioinformatics, Whole Genomes and the Future of the Field
Hello. This is Ann Nguyen, Senior Associate Conference Producer with Cambridge Healthtech Institute. We are here for a podcast for the Clinical Genomics conference at Bio-IT World Conference & Expo 2017, whose 15th anniversary is this year, and which runs this May 23-25 in Boston, Massachusetts.
Right now, we have one of our featured and frequent speakers, Dr. Liz Worthey, Faculty Investigator and Director of Software Development and Informatics at HudsonAlpha Institute for Biotechnology.
Liz, thank you for joining us.
Thank you for inviting me.
Can you describe recent advances in informatics as they pertain to genomic medicine?
I think one of the things that's important to state is that it's really advances in technology, IT, and informatics that's driving genomic medicine along, obviously with advances in the sequencing side of things. We're seeing a number of things coalescing at the same time so there's lots of improvements in terms of compute and the ability to store that data, but at the same time we're seeing advances, for instance, in the secondary analysis phase. Taking the data off the sequencer and producing variant falls. It's come down staggeringly fast from eight hours just a short time ago. Now you can do that step in less than an hour. That means you can do more of the analysis and it's become cheaper and allows us to analyze more genomes.
We're also seeing much more access to different sorts of data that you want to bring in when you're doing genomics for clinical purposes and particular things like Daniel MacArthur's allele frequency dataset, like ExAC. That's also really had a big impact on our ability to do genomic medicine.
I think also we're starting to turn the tide to having almost more of a systems biology approach. So systems biology was a dirty word maybe five years ago because we tried to do it before but we didn't have enough data or compute to really work on it. Now we're seeing people integrating lots of different sorts of -omics datasets into these sorts of analysis. That's only going to help us generate new knowledge.
I think probably the last thing would be there's a real focus moving towards software development and software development practices that are more akin to what we see in other fields. It used to be the case that informatics and biology was not always carried out or performed by hardcore developers. We're seeing a change there as well in terms of the things that are being designed. I think all that's going to happen at the same time.
Why is it important to focus on whole genomes for clinical purposes rather than just exomes or gene panels?
Let me just state by saying that a clinical genome is the best first test when you have a patient who's either undiagnosed or has a rare disease that's believed to be genetic in nature. There's lots of reasons for that. One is when you do a genome you actually get better coverage of the exonic part of the genome than you do if you do an exome. There are steps in the process of doing a panel or an exome which means there's dropout of certain regions that consistency is much more consistent when you do a genome and exome. It's also things like when you do a genome you can use that data to call structural variants much more easily and efficiently than you can with a panel dataset or an exome dataset.
One of the things that we're finding is that we were getting 4 of 5% more diagnosis when we can do structural variations in addition to small variant calls. For me those things are really great arguments, but always comes down unfortunately in some ways to cost. The truth of the matter is that people often say that you should do a panel or an exome and then reflex to a genome if you don't get an answer. That's actually a really expensive strategy.
Say you did an exome. In general you're only going to get a diagnosis from an exome of 35% of the time. That leaves 65% of those patients undiagnosed, and then you would go on and do a genome for those. That cost is significantly more than the cost of doing a clinical genome alone. We know that clinical genomes give twice as many diagnosis as clinical exomes. Even the cost an individual patient and the cost of the healthcare system, genomes first is cheaper clinically.
I mean it's different when we're about for research purposes. But for clinical purposes, it actually makes more sense financially to do a genome straight off the bat. When you combine that with the fact that you get more data and you make more diagnosis in general, for me it's kind of a no-brainer.
How can the community of bioinformatics experts, genomics researchers, and other collaborators move the field of genomic medicine ahead?
There's lots of different answers to this question. One of them would be let's stop reinventing the same wheel. Unfortunately the way that the system is set and the funding is set means that people often are developing tools that somebody else is developing. There are many people that are focused on the same part of this problem. We need to work at how to get these people to work together so that we can actually work on the next part of the problem and the part of the problem after that.
In addition, I think it's at least clear to me that we need to stop making tools that only an informatics expert can use. Oftentimes we see these pipelines where it's command line. Somebody has to log in and run it. What we actually do…is try and develop tools for clinicians to use because the clinicians are the people that are doing this work, the directors.
I think another part of it is we need to tackle some of these other parts of the problem. We know reasonably well how to get from sequencing data to prioritized variants, prioritized based on how deleterious they are. We still have a lot of work to do on the phenotype side, including to be honest, work to extract clinical information from EHRs. I think one of the things that we could do is have the software folks at EHR companies work with the people who are doing informatics in clinical labs also working with people who are experts in research datasets that can be used to improve that process and really to get all those folks working together to make tools that clinicians can use, that researchers can also use.
I think a lot of it is trying to work out how to fund people to work together in a way that they haven't really been working together currently, because they're really funded just to work on their little part of the problem. I'm not sure where that funding would come from, but it's definitely something that we need to think about.
Thank you for sharing some very valuable insights into your work, Liz.
Thank you very much.
That was Liz Worthey of HudsonAlpha. She'll be speaking during the Clinical Genomics track during a session that's shared with the Data Security track, all of which is happening at Bio-IT World this May 23-25 in Boston.
To learn more from her, visit www.Bio-ITworldExpo.com for registration info and enter the keycode “Podcast”.
This is Ann Nguyen. Thanks for listening.