Kashef Qaadri
PROLOGUE
Charlene Son Rigby:
They are really looking to get their test validated and launch their test as quickly as possible. Being able to do that in even as short as a three month period and being able to kind of set up there, you know, the NGS operations, um, as well as getting the IT systems piece up and running is a pretty significant set of activities to complete. The cloud really enables them to do that.
Kashef Qaadri:
My name is Kashef, and this is BioRad.io — a group of biologists turned bioinformaticians bring you into the world of research and development informatics by interviewing the people responsible for implementing systems and technologies to a unique and diverse set of use cases. Several biopharma organizations are dealing with similar issues, technical, executive buy-in, mindset. Well, organizations can learn from the experience of others and how they’ve dealt with some of these same issues and perhaps strategize in ways of moving to the cloud. Today, we’re here with Charlene Son Rigby.
Charlene Son Rigby:
Hi, I’m Charlene Son Rigby, the chief business officer at Fabric Genomics. I’ve spent the last 20 years in a combination of enterprise software and bioinformatics at both startups and Fortune 500 companies, working with customers to enable them to get actionable insights out of large data. I’m currently at Fabric Genomics where we deliver a software platform for highly scalable genomic analysis. We enable organizations to take raw data off of the sequencer and turn it into clinical insights that can be used for management of patients.
WHEN DOES CLOUD MIGRATION MAKE SENSE?
Kashef:
Thank you for the introduction and thank you for being here. Let’s jump right in. What do you think are the main motivations for an organization to move to the cloud, particularly now in, in this day and age?
Charlene Son Rigby:
So from my experience, organizations are seriously looking to move to the cloud, and moving to the cloud because of a number of factors. So in the genomics world, the size and the complexity of the data is growing. The cost of sequencing has continued to decrease, and as a result, data volumes are growing quite rapidly. So at Fabric, we work a lot with whole genome data.
And to just give you a sense of that size, a raw data file called a FASTQ file that will come off the sequencer is about 90 gigabytes. And typically you’ll have two of those data files per sample, so you’re talking about 180 gigabytes, and that, per patient, and that adds up very, very quickly to many terabytes. That’s one major issue. So handling that size and complexity of the data.
Kashef:
Sure. So just diving into that a bit more. So you’re talking about size complexity, that obviously creates a storage need. What are the other elements that are really important or motivations to move to the cloud outside of just the storage of those data, large datasets?
Charlene Son Rigby:
Yes. Another factor is lumpy data processing and growing data processing needs. So in next generation sequencing, data comes off the sequencer in batches. So you have this basically surge of a high data load that’s required for processing, but then that processing machine might sit idle.
And cloud computing provides on demand processing and it enables companies to utilize this lumpy data processing essentially to their economic benefit. And it also enables really seamless scaling as organizations start to grow their data loads over time. Cloud computing is also what enables software as a service companies, you know, including Fabric to price their products at more reasonable price points even with high-performance AI technologies that are involved.
Kashef:
Right. So you’re, you mentioned the lumpiness of the data. Is that referring to the streaming of the data directly from the instrument to the cloud, or are you talking about analysis pipelines and the sporadic nature of the hardware to perform those analyses?
Charlene Son Rigby:
Yeah. Great, that clarification. So it’s in genomic data processing, a lot of it is around the sporadic nature. Because if you, for instance get 96 samples or 350 samples off of the sequencer all at once, then those need to be processed as, kind of in parallel, as parallel as possible so that you can get to the results, analyzing the results as quickly as possible. And then after that large load of analytical processing, that machine might sit idle until the next batch comes off of the machine. So a cloud environment would enable one to only utilize that machine and incur that cost when that high data processing load is being done.
Kashef:
Right. So you’re talking about the, just being able to only spin up the requisite number of servers to perform a particular task, spin those down, and then move on to the next step, that might be less, less intense.
Charlene Son Rigby:
Right. Exactly. And software as a service providers can really kind of amortize that usage across multiple customers and really deliver economic value to their customers in that way.
Kashef:
Sure. So you mentioned two things. One is around performance and being able to scale up and down, the other is around cost. Are there any other sort of big motivations to move to the cloud?
Charlene Son Rigby:
Yeah, definitely. There’s a couple more that I’d mentioned. One is around teams. So we are seeing a lot more geographically dispersed teams. So an example of this is we might have geneticists that are performing interpretation in multiple office locations or multiple hospital locations, and they might have a particular disease expert that is in another state or another province and they were calling that disease expert into review a specific case.
The cloud and cloud solutions will enable that group to collaborate effectively and be able to work on the same data in their normal workflow, so not having to somehow extract that case, its specific case information and go and find a different method to be able to get that expert to look at that information and bring it back. So this is just through the normal analysis workflow and do it in a secure way.
Kashef:
Right. Yeah. And just to let our audience know, we have a separate topic on sharing and collaboration that we’ll get to in, in a few episodes. Just going back to your earlier, sort of requirements, kind of setting the stage for why organizations, biopharma or biotech companies, research organizations are moving to the cloud.
In the context of startups, you mentioned those, they typically don’t have the in-house hardware existing, right? Could you, could you speak to that in terms of the ramp time, going from zero to being able to perform analysis, comparing traditional or older pharma companies versus newer startups that may not have the necessary hardware?
Charlene Son Rigby:
Yeah. That, I think that’s a great point in… We have a number of labs, uh, regional commercial labs that are, they may have been doing other types of testing and they are transitioning over to doing next generation sequencing testing. They don’t necessarily have the 50% IT team that a large reference lab might have or a pharma company would have. And so they’re really looking to get their test validated and launch their tests as quickly as possible.
So being able to do that in a, even as short as a three-month period and, getting the equipment in, sourcing the sequencer, being able to kind of set up the NGS operations and going through their regulatory, as well as getting the IT systems piece up and running is a pretty significant set of activities for a regional commercial lab to complete. And so the cloud really enables them to do that and they know they are able to get the robust functionality they need and the workflows they need without having to build it themselves.
CLOUD-FIRST VS. LIFT-AND-SHIFT MIGRATION
Kashef:
Right. So you touched on two, two additional motivations, right? So the, the ramp up and speed at which a lab can go to production, right? Being able to set up their sequencers, but then also be able to supplement the, or create the necessarily analysis pipelines. And then the second point that you mentioned is a lower requirement for resources. And, and that’s huge. The cost, time and cost, around being able to maintain these servers, physical hardware on premises is quite expensive.
So earlier we talked about taking a cloud first approach, I guess that is more for startups that probably don’t have existing infrastructure as compared to a more opportunistic approach where here’s a new use case.
Maybe we’ve outgrown current infrastructure and need to scale up, or maybe this is a new type of data that we’re dealing with, and the requirements are quite different and don’t fit in with the on-premise servers and databases that we have. What do you think is the approach in terms of how customers prioritize which projects, which projects go to the cloud and how they sort of approach that?
Charlene Son Rigby:
Yeah, we’ve seen very different approaches. So I think it’s very mixed depending on the organization. Certainly new projects, it’s quite easy because there isn’t a legacy infrastructure and there’s an opportunity to take advantage of the newest technologies. So that’s an obvious one.
Others that we’ve seen are projects that are, have become for one reason or another difficult to manage because of the size and the complexity of the data or the compute, and so they become such a pain point that they need to be moved out of the internal infrastructures. And another thing we’ve seen is that projects where organizations are starting to experiment with the cloud, they will put forward projects where there’s a concern about data security.
And so putting forward projects where there is, there’s fewer security concerns about the data. So that might be the data’s, you know, totally de-identified or they’re, they’re specific types of research datasets. Those might go out to the cloud first before, before other types of data sets.
CLOUD SECURITY CONSIDERATIONS
Kashef:
So let’s unpack that a bit. What do you think about, what are your thoughts around cloud security? Is it really as secure as people advertise? What are the reservations and how do you overcome those reservations?
Charlene Son Rigby:
Well, data security I think generally is a very hot topic. You probably saw in the news earlier this week that there were 22 Texas towns that were affected by a ransomware issue. And I grew up in Texas, so that was very close to my heart.
Hackers, and attempts to get into IT systems are happening all the time. And so when organizations are thinking about a cloud environment, I think that there’s a growing realization that cloud providers can provide state-of-the-art security versus an organization needing to build all of those functions themselves. And so that’s a big driver to really, strongly considering and moving to the cloud.
Kashef:
Right. And do you think a customer at the end of the day, specifically in the context of sensitive data, do you think they are more or less vulnerable kind of looking at on-prem versus on the cloud approaches for storage?
Charlene Son Rigby:
From my viewpoint, a state-of-the-art cloud environment should provide a higher level of security than an internal environment because of the fact that this is a product that is being provided by a commercial provider, such as an AWS, such as a Microsoft. And so that really enables an organization to leverage those capabilities versus having to build it themselves.
Kashef:
Right. Excellent point. So, in addition to the security, I am sure you come across a lot of concerns, reservations of moving to the cloud or an organization implementing some sort of cloud environment. What do you think are the other big hurdles or reservations that a customer might have?
Charlene Son Rigby:
I think that the main reservations that we see are around control. And, you know, certainly there is a comfort level of having data within one’s own internal environment. And that’s kind of the way that we’ve been doing things, there’s a comfort level with that being the existing systems. And then the questions around, and concerns around data security, especially when you’re talking about patient data, identifiable data, those are, I would say, the top of mind concerns.
Kashef:
So it sounds like when you were developing your solution, you took a cloud first approach. I would imagine you didn’t build anything in house and then try to push it to the cloud. What do you think gets left behind when an organization just does a simple lift and shift?
Charlene Son Rigby:
Mm-hmm. So certainly the cloud does offer security, it offers a lot of capabilities that a lift and shift approach can benefit from. But the, some of the key benefits around being able to support lumpy data processing, being able to support short-term usage of very high performance machines and also being to scalably manage large, large amounts of storage with solutions that might include long-term storage that is priced very differently from short-term storage. Those would be some of the things that would not be, an organization would not be taking advantage of with a lift and shift.
STEPS TO CLOUD MIGRATION
Kashef:
Right. Absolutely. And I guess another approach to that would, would be refactoring, which is sort of a hybrid approach where you wouldn’t necessarily have started off in the cloud, but you’re adding some of the capabilities around the scalability and performance to take advantage, not just of the basics around security, but then also taking advantage of the sporadic nature of the analysis pipeline. So sort of shifting gears a bit, what are the steps that an organization would take taking this vision of moving to the cloud and making it a reality?
Charlene Son Rigby:
So I spent several years doing a lot of change management at a Fortune 500 company. And on some level moving to the cloud is actually quite simple because there are a number of readily available, very high quality cloud solution providers on the market. And moving to the cloud, they provide facilities, if you’re even talking about the lift and shift approach, it’s quite easy to move applications over and large stores of data over. I think that a key aspect of doing that, making that shift is really around the organization and change management. So getting buy-in from the key stakeholders and really, especially not only…
People believe that the IT organization is the organization that needs the most buy-in here, but it is also the business organization that really needs to buy-in to fully make the shift.
Kashef:
Right. And just speaking about that paradigm shift, trying to get that executive buy-in, there’s definitely a change in mindset. What advice or any recommendations that you would offer to the listeners in terms of trying to get that shift in mindset and approach?
Charlene Son Rigby:
I think that it comes down to the business case and really outlining what the business benefits are of moving to the cloud as well as addressing concerns around data security, which is again paramount for every organization and what is the change plan.
THE FUTURE OF CLOUD
Kashef:
So what do you think is next? Do you, do you see the cloud going away? Or I guess, how do you see organizations approaching the cloud people that are, organizations that have been hesitant to date, do you see that changing? Do you see that shift where things are moving more to the cloud or, or not?
Charlene Son Rigby:
We do see a trend of increasing adoption. So generally speaking, we saw that the US was an earlier adopter of cloud solutions over certain, certain other regions in the world. And we’re, we’ve been excited that we have seen growing adoption of cloud solutions internationally. What we see accompanying that is a desire to keep data in region.
So it’s not sufficient to have a single cloud server and serving the entirety of the world. And that’s understandable given local and regional data laws that organizations need to abide by and want to abide by. And so at the same time, the cloud provider capabilities have been growing. So AWS has many data centers around the globe and as does Microsoft and other providers. And so these are really critical enablers of cloud solutions. And so what we are able to do is basically deploy our software in multiple data centers and be able to address-
Kashef:
And in multiple regions.
Charlene Son Rigby:
Exactly. And to be able to address that need.
Kashef:
Sure. So you’re, you’re checking off the box in terms of localiza-, or regionalization of the data, making sure that it’s stored and processed locally. I guess that really speaks to the advantage of being able to deploy internationally through the cloud. One of the challenges I’m always thinking about is if you have data that are localized, so just say it’s European data, it stays in Europe.
What happens if that researcher wants to come to the States or collaborate with someone in a different localiza-, or in a different region? How does that work out? And, and I think there, there are some, additional challenges that we haven’t really thought about when it comes to cloud and, but these are good problems to have. If you’re working on premise, it doesn’t change how you approach those problems either.
Charlene Son Rigby:
Yeah, I agree. I, so a lot of the ability to data share in the industry that I work in is around, what’s in a patient consent, what is in an IRB and so those consents and those IRBs had to have been set up to enable that data sharing. And so that’s kind of a fundamental requirement. And then beyond that, as you said, how do we facilitate it? And in the example that I gave earlier around a lab needing to reach out to an expert in a particular disease area, them being able to do that in an environment where it’s in their normal workflow so they’re able to set that up as something that, it’s an exception that they need to ask the expert for their evaluation, but they’re able to do it as a normal course of their work. I think that that’s really important for the progression of patient care as well as for research.
Kashef:
Sure, absolutely. So thank you for listening to BioRad.io. I’d like to thank Charlene for being our guest today, talking about moving to the cloud. To join the conversation, visit our blog BioRad.io. And don’t forget to subscribe on Apple Podcasts.