Golden Helix Looks Ahead After Receiving NIH Grant For Structural Variation Detection

October 16, 2018

October 16, 2018 | When Andreas Scherer took over as the CEO and president of Golden Helix in April 2013, he knew it was time for the company to shift from a research focus to a more clinical focus. With his Ph.D. in computer science, a master bioinformatics, and international business experience, Scherer was uniquely qualified to implement Golden Helix’s expansion into the clinical space. The company launched its first clinical product in 2014 and a new product every six months or from then on out. This September, the Golden Helix celebrated its 20th anniversary and announced another grant from the National Institutes of Health to dig further into detecting structural variations.

Bio-IT World editor Allison Proffitt sat down with Scherer to discuss the company’s history, the shift in trajectory over the past five years, and what’s next. (This conversation has been edited for length and clarity.)

Bio-IT World: When you took over, Golden Helix shifted from more of a research focus to a clinical focus, and that's when the complete pipeline came together. Why did you think that shift was the right fit for where genomics was going and where the industry was going?

Andreas Scherer: After the Human Genome Project finished, the whole industry was primarily focused on finding any type of association between genes and diseases. This process continues, but over time, our ability as a community increased to conduct clinical analytics. Very early on we were very successful with developing tools in the research market. This is a great market in which we are still very active in, but also it has certain characteristics that you would not find in other business areas. There's a lot more change in research, and as a result, from a business perspective, it’s a little bit more demanding to keep the revenue levels up.

However, that's not the most important thing. The most important thing is as we gained more insights as a community into the relationships between genes and associated diseases. As a result, the applicability of that knowledge in the clinic rose. We already had customers that used our research-based tools in the clinic. We needed to shift the entire company to a clinical focus because that will be the future of this space.

When I took over, we did an inventory of our entire business and looked at our technology stack and then started right then, right there, pivoting into the clinical space. The first product came out the year after in 2014.

Were any of the same products brought over, or did they all have to be retooled for the clinical space?

We were able to leverage our underlying tech stack — specifically our data management capabilities.  This part of our stack is the foundation of every analytics product we launched. It allow us to implement products that scale and fast. We were able to build out new products based on a very robust, tested and proven stack. Obviously, we had to build out an entirely new set of products to meet the requirements of our clinical uses. So, unfortunately just retooling our existing research platform was not an option.

For us, this entire process was simpler than for any company that just started in this space, because we already had the technology stack underneath that was very thoroughly vetted. Also we had a significant amount of domain expertise. We understand this space in and out, and we have many employees that have been with us for a long time, so there's a lot of corporate knowledge that we were able to leverage in that process.

You rolled out your first product, the first clinical-focused product, in 2014.

Yes. We started with the first part in 2014. Every six months after that, we came out with a new solution that completed our tech stack. We started with VarSeq; then we launched VarSeq Reports. Then we released VSPipeline, a product that allows the automation of entire analytics workflows. After that, we created a genomic data warehouse product called VSWarehouse for larger labs. In this warehouse product, we essentially collect all artefacts of a bioinformatics pipeline and make those retrievable.

How does the data warehouse work? You're not in the business of storing data, right?

Right. It's the software product that makes the data accessible and retrievable. So the idea is, every artifact that a bioinformatics pipeline creates goes into this data warehouse. We don't sell the underlying server infrastructure. You can have the system either reside on your server or on an Amazon cloud our any other cloud service. The warehouse is the capability that allows a lab to look at every potential variant that they ever saw and indexed before.

Let's say you run a sample you find nothing. That is not uncommon. Moreover, so you have a disease, but you don’t have any diagnoses, yet. One year later, there is a new paper that comes out, and all of a sudden there is a correlation between a variant and a disease that had not been described before. You can re-run your entire patient pool, and if there was someone with that variant, it would pop up.

The other big use case is that most labs have multiple analysts, specialists that are looking at samples. You can create your own “assessment catalog” for your specialty in your lab. If you maybe focus on particular disease categories, you can create your specialized assessment of a variant in the context of the observed disease and what to do about it. For most labs that's the secret sauce if they have a specialization in a particular area. You can use the warehouse to apply this in a systematic way across the entire diagnostic process.

What is the latest addition to the pipeline?

We have launched a product that allows our customers to conduct CNV [copy number variation] analysis in NGS data. That is one of our key differentiators, and here it's starting to get interesting. The product that we have currently is already very, very good, and we got additional funding from the NIH on a smaller scale to get to that level. There are customers that have already completely omitted the use of confirmatory methods such microarrays or MLPA to spot and to detect CNVs.

They do this all just with our tool and next-gen sequencing data. It not only simplifies the clinical workflow, but it also saves much money. When you look at precision medicine and the big picture, this is only all happening if we come down in prices and make it affordable so that payers can actually agree to deploy these tests on a larger scale. So now our ability is very, very differentiated and probably market-leading when it comes to detecting these more substantial copy number variations straight up in next-gen sequencing data and it's very fast, too. High quality and fast.

In mid-September, Golden Helix announced a National Institutes of Health Phase 2 SBIR Grant:  “Integrating CNV analysis into a NextGen sequencing clinical analytics platform.” You already have a CNV analysis tool; how will the grant improve it?

The NIH has been a tremendous partner, and I can't tell you how grateful I am that we have been selected.

We want to deeper integrate the CNV analysis into clinical workflows based on guidelines, such as ACMG guidelines. Nobody else does this; that's a huge differentiation. Moreover, secondly, we are leaping and also looking at much rarer, much more difficult to spot, complex variations such as inversions and translocations and gene fusions. That's an area there where we are applying a lot of that research money, and so this will be on the roadmap for the next couple of years.

I'm the lead researcher and Principal Investigator on this grant. The new funding allows us to deeply integrate this into the clinical analysis, which is key. Moreover, we have to work with stakeholders in our industry because this is not standard today. The CNVs and any other complex structural variations are part of the human variation, so that has to be taken into account when you look at a sample from a clinical perspective. Deep integration into clinical workflows has to occur. However, we don’t want to stop there. We aiming to further improve our detection capabilities. There is still, from a technology perspective, more possible. We have now the ability to take that calculated risk and explore that further.

What’s the roadmap as you dive deeper into integrating a clinical analysis of structural variation?

We have the current ability to recognize a CNV—copy number variation. We will integrate that into clinical workflows, and we will develop and extend our software packages in that direction. That is the first step. Moreover, then parallel to that, we look at other types of structural variations. The three that I would name specifically are inversions, translocations, and gene fusions.

Some people believe the structural variations is the next crucial step for understanding how the genome dictates everything, and that SNP analysis is only scratching the surface. Do you think that's overstating?

No. It's correct. If a clinical lab solely focusses on SNP analysis, they are missing out on a large spectrum of human variation.I will say this: if you do cancer analytics or assess germline mutation today and you're not doing a thorough CNV analysis, you will miss the picture. CNV analysis is crucial to understand the full picture.

Five years into your tenure—and 20 years in the company's history—where is Golden Helix going?

For us as a company, there are three pillars to focus on. First, we continue to focus on innovation. Everything that you and I just talked about is just an example of how we are staying on top of the latest developments: Pushing the state of the art of key capabilities such as CNV analysis capabilities along with our ability also to obtain external funding to develop cutting-edge technologies. We have been recognized as an innovative leader in multiple news outlets over the last few years

The second key pillar for my company is a strong focus on quality. This is for our customers very important. They get a high-quality product from us that they can rely on. Ultimately they make decisions that affect patient outcome.

The third pillar is about our relentless focus on customer satisfaction. One of the key differences that we have in this industry is that we don't have any Venture Capitalists as a shareholder. I operate Golden Helix as a profitable and growing business. We have been able to basically bootstrap Golden Helix because we have a relentless focus on keeping our customers satisfied. Moreover, so they stay with us, they renew with us, and that's the fundamental concept that will be for us crucial no matter how far you look into the future.

What are you seeing in the space that is coming that needs to change—besides just CNVs to understand cancer?

From a tech perspective, there are a few areas that we are very closely monitoring at and will continue to explore. I would say that at some point Artificial Intelligence will be, in this area, inescapable. We already employ AI technologies within our products today. We will continue to push the boundaries in this area.  Also, we will continue to with our relentless pursuit to improve scalability and speed of our products. The amount of data that we're going to process of the future will be magnitudes higher to what we see today. The companies that will be successful in this field will be the ones who can build massively scalable capabilities for the marketplace.

Editor’s Note: This article was also featured on Diagnostics World.