Reducing the Complexity of Multiplex PCR

Full Transcript:

Q: Hi and welcome to the Radian Partners innovation series. The innovation series is really focused on highlighting emerging technology companies, particularly those coming out of Southeast Michigan. Almost about two months ago, we had the CEO of DNA Software on this video cast series to talk about the origin of the company, the premise of the technology and how it was being used today. Today I’m joined by Joe Johnson, vice president of business development at DNA Software. And we’re going to drill a little bit deeper into the technology, its value proposition, its benefits, and we hope to illuminate some of the tried and true benefits of it today. So Joe, welcome to the video cast.

Joe Johnson:

Thanks for having me, Mo. I appreciate being here. I’ve enjoyed watching your innovation series and thanks for including DNA Software and our ideas and sort of what we think we have to offer this problem of diagnostic design.

Q:Absolutely. So, Joe, I think one question that I’d like to start with is demystifying what you guys at DNA Software call PCR testing in particular, multiplex. This sort of methodology, this capability, what is multiplex really about?

Joe Johnson:

Yeah. Good question. So multiplex PCR testing is important for several reasons. Namely a multiplex, some people might not know the terminology if you think multiplex, multi means many, even many years ago you’d go to the multiplex, which is a cinema that would have 20 screens or something. So a multiplex is the ability to do many tests at once. In in our case we use multiplex designs to give more information from the same sample well. So for example, if you have a multiplex COVID test, you would be able to put many targets into the same sample well instead of running those tests individually and separately, that takes additional time, resources, takes additional money. It also takes additional samples. So sometimes in some cases you might not have as much patient sample to test, but a multiplex PCR will give you multiple targets with sort of reporting.

Joe Johnson:

Let’s say for example, not just COVID-19 but also the variants, the UK variant, the Brazil, South African variant, DNA Software would be able to design all those targets in the same sample well and then report more information. That’s what multiplex PCR is known for is to sort of squeeze out of that same diagnostics, so to speak. So right now we’re actually working with a partner to develop a COVID variant diagnostic in a multiplex. And we’re also getting ready for the next, not just the ones you read about in the news or actually these are in the news also, but California and New York variants are of particular importance and what we’re trying to do is provide more information from the same sample, so to speak. So that saves time, that saves money and it provides ultimately more information from the same sample well.

Q:

Mm-hmm (affirmative). And Joe in the absence of multiplex PCR, it seems like the reason that there is a need to create the sort of operational efficiency, save time, et cetera, is that in sort of a singleplex world there’s a lot of trial and error. There’s a lot of redundancies going on. So talk a little bit more about this. What is this experience like for, let’s say a group of scientists trying to use trial and error to sort of navigate a singleplex capability relative to multiplex and how much time does it really save?

Joe Johnson:

Yeah. So oftentimes what I’ve learned in over 10 years DNA Software is that scientists will take sort of an experimental trial and error approach to designing PCRs and there’s limitations to sort of how well that works. Namely they use free software and there’s limitations on the free software. In fact, John SantaLucia our company founder, who you had two months ago, it’s the information that he published in a literature, those parameters are in the free software that’s used by some of the free tools. Now, if you’re designing a singleplex that’s rather pedestrian and isn’t particularly challenging you might get the design to work, but if you design a multiplex and you introduce more complex targets and more variables, there’s going to be limits to what free can resolve in terms of performance of the assay, in terms of it just functioning at all.

Joe Johnson:

So I think scientists work with the tools that are available to them. And what we’re trying to do is bring greater awareness to there’s better tools available. There’s a company like DNA Software that can work directly with R&D organizations to provide these multiplex designs so that… Like we’ve talked about, Mo, this is not necessarily a part of the R&D life cycle that organizations enjoy so to speak, multiplex is sort of a means to an end. They want more information. They want to improve the quality of the test and test for multiple targets at the same time, but the process of designing that is very complex and it’s oftentimes frustrating. I’ve heard from 100s of scientists in my time at DNA Software and they’ve shared just how difficult this can be and how frustrating it can be when it doesn’t work.

Q:

It’s so interesting, Joe, that solving a problem that’s so complex, that requires so much depth scientifically but also computationally, et cetera, that the sort of defacto solution for it tends to be an open source solution, that it tends to be something that is free or maybe you might even consider rudimentary in certain ways. Of course, the world has sort of decided that open source technologies, free software, it serves a purpose. And that purpose usually is that it’s used as a foundation to build upon that. But what’s so interesting about the PCR design world is that the free technologies are not actually being used to serve as some kind of foundation, that whether it’s a pharma company, a diagnostics company or even one of the interesting genomics companies in the space, they’re not using it as simply a building block to a bigger solution that they want to customize.

Q:

They’re simply using off-the-shelf open source technology as the total solution for their PCR design needs. And it’s just surprising to me again, given the complexity of what’s required and given the important of a lot of the problems that PCR as a methodology is solving, whether it’s the development of new vaccines or new drugs.

Joe Johnson:

Yeah. So, it is. It’s interesting that these organizations rely on free software to perform a mission critical function. It’s not a perfect analogy, but a lot of times organizations will use Excel, which is not free you pay for a subscription to it. But there’s ways people use it and bend it and build these macros that are huge and then if something fails or if you inherit a sheet that has an issue it’s prone to error, the more complex function you’re trying to perform with that free tool, the more quickly you’re going to find out the limitations of it. So we’ve learned just from the customers we serve and examples of when they’ve kind of taken a free tool to its limit and learned the hard way that, okay, now we can’t get this difficult target designed or that we can’t get it to perform in multiplex.

Joe Johnson:

And we don’t discredit the free software. There’s a time and place, and it has its function. But if you’re trying to, certainly in the case of multiplex or introduce some more advanced applications, you’re going to sort of find that limit so to speak with free software. So part of what we’re doing is bringing awareness to the community that we have best-in-class multiplex tools, we were awarded that last year by Frost & Sullivan. Not only do we have these solutions, we partner with organizations to perform a function that isn’t always something that they enjoy or really want to invest in. Sometimes this is just a part of the process, and we’ve got a lot of expertise. We’ve been doing this 20 years. We thought about multiplex quite a bit so that that organization doesn’t have to.

Q:

Mm-hmm (affirmative). One of the capabilities Joe, that you talk a lot about with respect to DNA Softwares tools is the automation capability set, especially within a multiplex kind of context. Obviously many of the open source tools and free tools that exist out there when it comes to this kind of robust functionality, they may not necessarily have that, but for DNA Software in the context of automation and in the context of testing different variants, how does this really work? And what kind of value proposition does automation have for scientists that are trying to solve this PCR design problem?

Joe Johnson:

That’s a great question. So DNA Software migrated its solutions to the cloud almost a decade ago. And we were kind of early, quite honestly in taking our algorithms and models and then leveraging cloud computing to give sort of better solutions that desktop computing simply couldn’t. There was limitations on memory. And although our algorithms and models were very sophisticated and robust, there was limitations based on 5,000 sequences at a time and working around the sort of memory limits of desktop computing. So initially, there was some pushback from our customers and some skepticism around taking proprietary sequence information and then putting that over the cloud. But that skepticism quickly relented when they realized that the only way to get these valuable designs and that the only way to sort of provide these highly sensitive and specific designs was to leverage cloud computing in these, what we call databases, well, play lists rather, we borrowed the idea from Apple, this playlist that we’ll include in your design, both sort of in the front end to sort of detect what you want. We call it an inclusivity database or playlist.

Joe Johnson:

And then in the background will be the entire human genome and even more than that, the microbiome or the entire reference. So using these playlist, you can leverage cloud computing and the quality of the designs can only be created, so to speak from using cloud computing. So we have this capability to automate with these playlists that we stack together and we have some algorithms like thermal blast that scan the entire human genome for any off target effects. And then as it relates to the COVID variants, this is interesting because now we sort of started off the conversation around multiplex and how a multiplex result could give you more information about what’s in that COVID variant. Now that could inform things like treatment or a vaccine escape or other epidemiological responses, things that might not be known if you just gave a standard COVID test or if you relied on sequencing, which can take several days, a qPCR result could be a few hours and have all this information present.

Q:

It’s so interesting to me, Joe, that for so many years in Ann Arbor, Michigan, there was an innovative company solving a problem in an area that probably not many people understood for years and years and years. But now, with the COVID environment and with all the effort that’s been put into the development of vaccines and also related solutions on the biopharma side, it seems like what DNA Software is doing is more relevant than almost anything else I can think of in this kind of COVID environment. We’re still as a country, as a globe, we’re trying to solve this problem. What is the experience been like in talking to customers, whether they’re again, biopharma companies, CROs, diagnostics companies, genomics companies that you’re working with, has there been sort of an uptick in interest in DNA Software?

Joe Johnson:

Absolutely. There’s been not only an uptick, an appreciation for what we’ve done and built. Overall there’s been sort of an increase in the reliance upon DNA Software, not only to provide these customers… I think we’ve had about 30 organizations use our software for some type of COVID diagnostic. We won this award actually from Amazon, this Diagnostic Development Initiative as a result of our service to the community for these companies that are producing COVID diagnostics. But beyond just the COVID diagnostic there’s been even an increase in multiplex panels for respiratory disease. So one thing that’s been really important is knowing, is it COVID? And if is not COVID, is it flu A or flu B or some virus that’s in this list of respiratory organisms? So that’s been an important sort of recurring project for us.

Joe Johnson:

We’ve worked with about 10 different organizations, I’d say in the last six months to deliver these types of panels. So from a customer perspective, it’s never been busier. And just from a personals perspective, it’s funny because about over a year ago I would explain what I do to people and their eyes would kind of… I coach soccer and basketball for my kids. And one of the dads would say, “Yeah, I work at Ford. What do you do?” And I would try to explain what I do and I’d try to do it succinctly but I could see the person’s eyes slowly glazing over as I explained, but now you see PCR on the ticker on the bottom of ESPN. You hear the news every day. And when you tell someone, “Yeah, we have software to make PCR tests for COVID.” They get it. And so that’s interesting that unfortunately we’ve had COVID change our world and our landscape, but we have been working on this problem of PCR design for two decades so we are uniquely positioned to help companies that are trying to design PCRs, and particularly those that are doing multiplex PCR.

Q:

Mm-hmm (affirmative). Let’s talk a little bit about DNA Software’s data capability. It amazes me that there can exist technology that routinely is scanning data sources as vast as the human genome, in this case, in the context of detecting false positives and things like that. It just seems mind-boggling to me, I mean, how large that data set is and how much computational firepower you need to have to do that on sort of a routine basis. So, how does this work, I mean, in terms of DNA Software scouring the entire human genome to look for false positives, things like that, how is it powered? How does it work? How does it come to life?

Joe Johnson:

So it’s a combination of computational firepower, but also the way these algorithms have been built in correctly scoring mismatches, which is, I want to get too technical too quick, but in biology we learned A pairs with T and C with G. And when I first arrived at DNA Software that was the context of base pairing that I understood, but the reality is mismatches are not created equal and some are stable and some are unstable. And our ability to scan for those mismatches and then report ones that are stable or not is really what sets us apart. I’ve heard John say this before, we almost have built the Google, if you will, of search engines for matching DNA or mismatches to DNA. So if you think about the power of that, well, you can find right away if something, it hits or doesn’t in the quality of the hit.

Joe Johnson:

So the ability to score that well early on is important, but once you have these more robust and rigorous metrics for scoring, now you can bring the computational power to the problem. And that’s like we talked about, over 10 years ago when we brought cloud computing in to leverage that in combination with the correct scoring, now you’ve got something really powerful.

Q:

Mm-hmm (affirmative). Joe, I remember when I first learned about DNA Software, one of the things that I did is I looked at just the history from an academic perspective of this space, PCR design and how it has evolved over the years. And you look at literature from the ’80s, you looking at literature from the ’90s, you look at literature from nowadays, it’s interesting, you see similar names, similar methodologies, similar processes, et cetera. And it’s fascinating that from a methodological standpoint, we are now at a point where companies like DNA Software can really chart the future of what this category is. But still, scientists are really, really frustrated about this problem of PCR design. Beyond some of the issues that we’ve spoken about, just put us in a day in the life of a scientist who’s trying to solve this problem. Why is it so frustrating for them?

Joe Johnson:

Well, just from personal experience I think when you’re working on something you don’t understand why it’s not working, that can be frustrating. So PCR design is multi-variable, it’s multifactorial, there’s lots of nuance that goes into it and this might be oversimplifying, but if you’re working with a tool that’s really not built for the job, for example, this is maybe something we can all relate to but if you’ve ever bought in furniture from IKEA, you put it together with that three Allen wrench they give you, that can be infuriating. I mean, it’s not free. You paid for the Allen wrench with the furniture, but you’re using something that can get the job done, maybe, but you’re doing it in a way that’s suboptimal. All right. So I think because of they’ve learned through their graduate training in the literature that these are the tools that are accepted in which I understand, if you spent that amount of time in graduate school and then a postdoc and you were taught this methodology, you’re probably not going to deviate from that.

Joe Johnson:

So, that’s sort of the common best practice. But there’s sort of a lack of awareness that, okay, there are other tools available and there’s other strategies. Quite honestly, not only is it our design metrics and our calculations in our models, but it’s also how we go about some things that are different than what’s in the literature. Some of that’s proprietary, but some of it’s a lot of folks, this is going to be technical, but they focus on TM. Whereas percent bound is really a key metric, but it’s not in the literature of the way some other things are. There are lot of people do sequence alignments, but we actually look for target accessibility and complexity, which is to boil that down we’re going to find the sweet spot of where to design the PCR in a way that if you follow the herd in literature, you might not get there because of just the findings we’ve come across over the last 20 years in the space.

Joe Johnson:

So I think that the essence of why the frustration comes in is because they’re using tools that are suboptimal. They don’t know why the results are the way they are. And certainly in the case of multiplex, if they design a fiveplex they might get four of them to work but then every time they introduce a new target, one falls out. John, when he does his webinar, he uses the example of a whack-a-mole. So every time you think you have the problem solved another mole pops up. And that I think is a day in the life of a scientist trying to design a multiplex, using this tool and trying to wrangle a problem that they don’t necessarily have all the appropriate tools for. Yeah.

Q:

Mm-hmm (affirmative). The three words that anyone in the PCR design space hears over and over and over again, my guess would be are sensitivity, specificity, and coverage. I think those are the ones, right?

Joe Johnson:

Right. Very good.

Q:

And so I want to ask a question about the future, and this will be our last question. If that has been the focus of PCR design over the past several decades, what does the future of PCR design look like? What kinds of technologies, methodologies, approaches, capabilities are really going to fuel the next chapter of PCR design as companies like DNA Software lead the way?

Joe Johnson:

That’s a great question. So I think sensitivity, specificity, and coverage in the context of the multiplex. So I think there’s perceived limitations around those three facets in a multiplex. I think scientists often believe that, oh, this is not possible. For example, even the COVID variant that we started the conversation with, a lot of that’s done by sequencing because they think this can’t be done necessarily by PCR or that there’s going to be inherent limits. So I think the future is multiplexing. I think the trend of the economic benefits, the time, for example, we’ve done several respiratory panels that have 30 targets and you’re going to simultaneously report all 30 on a matter of hours sometimes maybe less, depending on the platform. And can you maintain sensitivity, specificity and coverage while there’s 30 targets present? Because that’s quite a trick. So we’ve spent a considerable amount of time working with customers to be able to pull that rabbit out of the hat and we’ve been able to do it.

Q:

Mm-hmm (affirmative). Well, there’s no doubt in my mind that DNA Software will continue to be a pioneer in this space. And more and more folks, again, whether you’re on the biopharma side or the CRO side or the genomic side will continue to partner with DNA Software. It’s always a pleasure to talk to folks from the company and learn about the latest and greatest. Joe, thanks so much for your time.

Joe Johnson:

Likewise Mo, thanks for having me.