Software systems to innovate and grow

Episode 49

Challenges, Success and the Future of the Biotech Industry with ArcherDX: Doug Wendel with ArcherDX

Show Details

Doug Wendel joins us to discuss ArcherDx, a pioneer developer of genetic and molecular laboratory technology. During this episode we hear a touching story of how the technology at ArcherDx helped doctors to make a medical suggestion, extending the life of a young girl with cancer.

Though the technology has not created a cure, this revolutionary moment in science shows the possibilities of where medicine is headed. Join us as Doug shares the many challenges of and the skyrocketing need for the biotech industry to continue growing and developing technologies.

Topics we discuss:

1. ArcherDX
2. Biotech industry
3. Genome sequencing
4. Genetic and molecular laboratory technologies
5. Fight cancer

Related Links:

1. ArcherDX
2. Jobs
3. Twitter
4. LinkedIn
5. Rosalind
6. “Biotech breakthrough eases Lafayette girl’s long cancer battle”

Show Transcript

Carlos: Doug, thank you so much for joining me today. How have you been? It’s been a while since we chatted that first time.

Doug: Yeah, thanks, Carlos. It has been a while. Definitely nice to be on. I was just mentioning, my wife is extremely busy at work but that feels pretty normal these days. So I guess I’ve been pretty normal.

Carlos: And I think everybody can relate. Everybody who is listening understands the challenges that we have in engineering. Engineering is the core of many businesses, let’s say it’s like the actual core to core production. Sometimes adding to the top line revenue for the company. What’s interesting is, for example, what we’re going to talk about today; There’s more than that. There’s more to life than redesigning the like button, some might say. And then there is also more to life than just top line revenue, which is very important by the way. But there is more impact that we can create. That’s why I am very interested and excited about today’s conversation.

In fact, let me do a brief introduction. So when you and I first connected I wanted to know a bit more about kind of the technical side of what you guys do at Archer. And just to confirm should we call it Archer or ArcherDX, the full name?

Doug: ArcherDX, it’s the full name.

Carlos: So I wanted to know exactly what you guys do at ArcherDX. In our last conversation, it naturally became an overview of this molecular pathology and genetic laboratories. To try to understand how genomic sequencing can actually help the fight against cancer with personalized molecular detection tools which is what you guys do at ArcherDX? But of course how does software and technology play a role in this big mission.

So anyways, I think another valuable lesson learned from this episode is that we’re trying to draw a map to explain how a genetic or molecular lab that does diagnosis works, right? Things like the essay, the sequencer, what do they do? You know, what are some of these repeatable components that exist in labs and how the systems, like the actual engineering systems, play a role to fulfill this role. So that’s kind of our topic for today. It’s a bit of a mouthful so I hope you guys enjoy this conversation.

So Doug just to get started, tell me a little bit about yourself. I wanted to get to know the man behind the story, so how did you get into tech? What drew you to this industry?

Doug: Yeah, that’s a good question, Carlos. I think, when I was super young, I was not sure if I really knew exactly where I was headed but you know my parents told me that when I was a kid I used to take part all of my toys. I would only sometimes successfully put them back together but I think that was sort of the first moment when maybe I had an idea and I’ve heard some idea that I was probably headed into some engineering discipline. And then, you know, once I got into junior high and high school, I think I had a pretty decent idea that I was really excited about computers and got into computer gaming. And shortly after I got out of high school I was sure that I want to be a software developer.

Carlos: What drew you to, you know… I know that at some point once you leave school, you have so many choices, right? There is something I always say that some people might actually find a little cheesy, people that I work with, I say that it’s easy to find work in our industry in the software industry as a whole but then we have so many places to pick from, right? Not only are the variables the actual workplace but the people that we work with. It’s always hard to find a good fit for us. So how did you end up on ArcherDX? What interested you to specialize in this field?

Doug: Yeah, I think some of it was partly deliberate and some of it was in an accident as well. I mean, I think as software engineers we wind up working in the industry early on, or maybe it’s some place that we find interesting or maybe it’s just the place that we got a job. You know, that’s definitely the case for me as well. My first actual tech job was, even before I started college, I wound up working in a networked PC gaming arcade which is both really fun but also taught me some pretty crucial skills. I did some web development there and some network engineering. That’s a pretty interesting group of people. Actually a side note on that, Jeff Atwood, is the co-founder of Stack Overflow. We were all big Quake and WarCraft players for quite a while. But anyway that was kind of my introduction to doing technology work and from there I wound up at the University of Colorado, Boulder in the Computer Science Department and just because I’ve worked at that previous job I was able to get another job in the SU Engineering Dean’s Office doing whole bunch of database work, web development, and networking and things like that. I think that’s kind of got me started on my path. You know, like I am generally interested in computer science too. I think I want to be doing full stack development. You know, I love seeing a product from start to end being able to do the backend work and then present the user interface to people get to work with.

And then from there you know there is a whole bunch of other things in between. I worked as a contractor for quite a while and really enjoyed that but ultimately realized that I appreciated the day to day interaction with people so I wound up taking another full time job at an aerospace software company. Worked there for quite a while and then. There is actually a former database professor of mine that sort of contacted me out of the blue and said, “Hey, there is this really interesting opportunity available at CU in Boulder, here is the deal.” They say it is a job working with Dr. Rob Knight. He is a PI or the Principal Investigator who is a brilliant and very well known microbiologist. He had a position open up and decided, “You know what this is so different from what I used to do but also so interesting and exciting.” Like what the heck I am just going to apply for it. So I did and you know it took about six months to actually land that particular job but I did. I think that’s kind of what kind of started me on maybe where I am in my current phase of my career. It’s is the first job I had where I really realized I am not just writing software but I am actually writing software for some purpose that fills intrinsically meaningful to me. The other really interesting thing about working at the University, where most companies are proprietary, the software you develop is sort of held closely to the chest but because we are mostly grant funded everything we did we shared with the academic community. So that was like a big sort of paradigm shift in my brain. Also when I took this particular job I had done almost exclusively Windows development until I started there, so you know, I was doing a lot of .Net and C# and some Java coding and things like that and certainly working in Windows environment and this was like a complete switch into MacOS and Linux, Python, Oracle and all these technologies I hadn’t used before. Plus a domain of knowledge that I didn’t really know anything about until I started. Yeah, a whole bunch of challenges for sure but I think that’s what made it feel particularly rewarding. I knew that I could come in and I knew that I could contribute right away. With my software engineering and my particular database skill set, it bought me the time I needed to really get up to speed with what the lab was doing as far as microbiology. You know, what are all these new terms, how does all this stuff work. You know, drinking from the proverbial fire hose pretty much every day but really enjoying it.

Carlos: One question, so I just want to set the stage for a bit of that shock. What is it that we’re doing? I mean just to get an idea of how we play a role in this industry? This is a very cheesy thing to say but, how are we helping humanity? That’s the thing that you’re actually making a big impact on. To quote ArcherDX’s website, which is basically a global fight against cancer using a personalized molecular tool. What does that mean? Tell me a little bit about let’s say, somebody gets cancer and they need to… in a way, fortunately, we now have better tools to fight it. We are very far from solving cancer and being able to cure all cancers but how does ArcherDX support that?

Doug: Yeah, that’s a really good question and certainly is core to what ArcherDX is doing. You know from a technical perspective what’s happening is that we’re transitioning from this world of looking at tissue samples under a microscope and literally looking at things that look weird or counting as best we can what we think the variations might be at very sort of growth or very high levels scale. We are transitioning from that into per molecule resolution of exactly what the DNA and RNA looks like within your body.

In the future, I think that the goal of every diagnosis should be to identify some effective therapy that specifically targets whatever form of cancer that patient has. But in order to get there, we have to have the tools to see exactly what happened at that molecular level. You know, some particular drug or therapy could exist but you won’t know what its effect is until you know exactly what happened in that patient and whether that drug can target the pathway that the cancer exhibits.

Carlos: Let’s take this in for a second. Let’s look at the history of this for a second just to get an idea of where we are. It’s 2018 and we are recording this. It’s 2018 but maybe listeners listen to this in the future. Let’s say that in the early 2000s we didn’t have the technology to even dream about this right? Tell me a little bit about how far we’ve come. You know, when did it get started in the early 2000s and how far have we come today in the last 17/18 years?

Doug: Yeah, for sure, so like you said, these tools did exist 20 years ago but they were extremely expensive to use. Maybe, don’t quote me on this actual figure, I think that the human genome project was on the order of about a billion dollars or something like that to actually sequence that first human genome. But you know around like in the very early 2000s it was about $100 million to sequence a single genome so that would be like you or me or pick somebody off the street. If you wanted to see what their human genome looked like it was going to cost you about $100 million. So that’s early 2000s and by like mid to late 2000s that dropped by an order of magnitude to about $10 million. It’s was a lot cheaper but still way more expensive than it is now. They could never be practical on a day to day scale. And then you know moving on in time like 2010 or 2011 it’s about $10,000 to sequence a genome. And then you know here we are today in 2018, we’re down, certainly under $1,000 to sequence an entire genome.

And realistically what’s also happening is that typically when you’re looking for cancers you’re not using what is called the whole genome sequencing or WGS. We are doing what is called targeted sequencing in which we are looking at a much smaller section of the human genome. What that allows you to do is run more samples on a single sequencer run and that makes it just cheaper in general to run. So you know, realistically at this point we are talking probably at $100 or couple of $100 per a sample, and even for targeted cancer sequencing.

Carlos: Alright, for everybody listening just bear with us. We’re going to go into some definitions here in a second because I know that you’re going to need to understand what a sequencer is to even understand the impact that has. So Doug will help us get there. Do bear with us for a second. From about $100 million to under a $1,000 in less than 20 years.

Doug: That’s pretty remarkable isn’t it?

Carlos: Not only that, it creates a different problem, right? Bioinformatics is an industry kind of born out of that problem, isn’t it? I mean the amount of data let’s say that comes out of the sequencers is tremendous and maybe you can give us a bit of actual size of in terabytes or something. But what sort of impact does it create in terms of now thinking of an engineering organization to support this sort of load.

Doug: Yeah, so you could imagine when it was $100 million to sequence a genome there was not a whole lot of those rolling around and so the software tools didn’t need to be nearly as robust as they do today. But to answer the question of does is a typical sequencing run look like in terms of how much data is coming out of it, I would say our typical essays or a typical sequencing run for a single sample is probably in the order of around like 2 to 10 million reads for one person. So that results in a fair amount of data to look through and that is just one person. So what has happened is that the cost of doing the sequencing is relatively inexpensive. In fact I would say it’s almost a non cost when you look at that in respect to or in relation to what the cost of developing the software pipelines and what is the cost of the actual computation is far more expensive now than the actual sequencing is. I think the problem is the domain has shifted from how do we generate a genome to how can we reasonably process hundreds or thousands of genomes in a single day and get accurate results.

Carlos: Just for the sake of making sure we’re understanding our terms. What is a read?

Doug: So that specifically means a single strand of DNA is a read by the gene sequencer. Typically what we are looking at with Archer is our reads or DNA molecules are about 150 bases long on either side. So that means it’s literally a string of molecules of 150 molecules long, which go into the gene sequencer. The gene sequencer can read that and tells us what molecules exactly were in which order and then we get a text representation of that the other end. So for every single one of those reads we get a few lines of text describing what is it in terms of As, Cs, Ts and Gs; those are the molecules that make up a DNA. And for each one of those samples imagine that we get, let’s say it’s 5 million of those for a particular sample and there are also two files per sample so double that it’s about 10 million different lines that we’re interested in for a normal sample.

Carlos: In our first conversation, you spoke about a very sad story but in a way a very powerful story about again the impact that we could have in somebody’s life here. I want to go specifically to that example, so you mentioned that at ArcherDX a couple of years ago… Tell us the story about this patient that came to you guys try to help show how somebody in the past couldn’t have even been helped, but somehow ArcherDX was able to step in and recognize something. Again, it was a sad story but it tells us about the power of the future ahead of us.

Doug: Yeah, definitely it’s a bittersweet story. Yeah, this is a story about a little girl. Her name is Zeta Matson. She was diagnosed with cancer when she was about 3 years old. She spent the next 9 years going through chemo, radiation, other drug therapies. She also had multiple surgeries to remove tumors and she had a complete hysterectomy and splenectomy and her other tissues removed from her body. You could imagine that poor girl is living with cancer for 9 years and has been through quite a bit of stuff that a normal girl would never have to go through. But you know, by 2014, it seemed that she might be actually winning the battle. Then in 2015 a new tumor was found that was not responding to any of those previous treatments. So her tumor DNA was sequenced, but it didn’t reveal any new treatment options. Then one of the guys at our company, a former employee of Archer, had happened to see something in the story. I think maybe his wife who sought the story, probably on Facebook, but somewhere on social media that story surfaced. And so Archer reached out and said, “Hey, you know we may find something, or we may not. If you can get us the slides or really just can you get us a tissue sample. We’ll run it and see what we can find.”

Maybe just a partial aside here. One thing that Archer can do that a lot of other companies can’t do is that we have a particular technology that lets us find genomic rearrangements without having to know all of the genomic rearrangements. So in most cases you need to know both ends of a change. But in Archer’s case you only need know one end of that change and we can find whatever happens to be on the other side of that particular rearrangement.

Back to the Zeta story, we got the slide. We were able to sequence it and in fact what we found was something that was a novel gene fusion that hadn’t been seen before. It’s a gene AKT1 was the gene. It’s a known meaning that it is implicated in cancers but it hadn’t been seen as a gene fusion which is a very large rearrangement of genetic material. So we were able to actually find that and then you know through some subsequent analysis and some partnerships with some other local and non-local companies we discovered that there are in fact clinical trials for these particular drugs but unfortunately they were all offered for adults. So you know, Zeta was too young.

However, we were able to, again it’s all through partnerships and good relationships we’re able to do this through MSK, able to get an exception for one of the clinical trials so that Zeta could go on this drug. So she started that drug I think in 2016 and it took about three weeks and most of that tumor has just disappeared from her body. So just kind of shows you the power of these types of diagnosis if you can understand exactly what it is and if there is a therapy for it. It can basically wipeout the cancer in a very short period of time. The really unfortunate part of the story is that Zeta did finally pass away in early 2017. Probably what happened here is that Zeta had had cancer for 10 years, and what happens with cancer is that they’re constantly reproducing and every time those cells reproduce there is a chance that one of those new cells isn’t going to look exactly like the cells that came before it. And so over a long period of time you wind up with not just one or two populations of cancer cells. You wind up with many different types of cancer cells. You know, some of which are going to respond to treatment and some of which won’t and unfortunately that’s probably what ultimately wound up getting Zeta. As this new tumor appeared it was resistant to the medication. You know, the new medication that was in that clinical trial and unfortunately she finally passed away.

Carlos: You know this is as you say a bittersweet story, but one that speaks greatly of ArcherDX and how you guys stepped in and we’re able to lend in hand in a time of need. But also gives us a little bit of a story of what the future might look like for future cancer patients. Unfortunately this is something that we as a society have to face. The way I think about this is that what more do you want to be motivated by, right? Because it allows us engineers to kind of team up together and try to fight this thing. So hopefully more people will come into the industry and I think that’s why one of the things that we’re going to talk about toward the end of the interview is how can we also motivate these people that might think that they need to be a biologist or geneticist to be part of this fight.

So before we talk about that though I want to talk about a bit more about ArcherDX. I want you to understand the role of the company and how it plays within the market. I think once you explain kind of the role in the market we then can talk about how to let other genetic and molecular diagnostic laboratories work with ArcherDX for multiple patients and so forth. Because I know you guys are a lab but also you work with other labs. I want to get a bit of that context of how you perform in the market.

Doug: Yeah, for sure and just a point of clarification we’re not exactly a lab. We don’t really do anything beyond our own R&D work. We don’t really do sequencing for the most part, for the people. But what we do is we produce all of the essays. This is literally a kit or an essay that we produce as an inbox of components that somebody will buy, that contains all of the necessary chemicals and reagents and primers, and I’m sure we’ll look into what that means in a minute. But it contains all the necessary parts for you to actually purify a tissue sample. Breaking that down to only the DNA and RNA components, amplifying the regions that we’re interested in and basically getting that ready to go on the gene sequencer. So you know, that’s like the physical product that Archer produces. That could be the purchasers of that product could be anybody from big cancer centers, to research hospitals to anything in between. Archer is certainly heavily in the business of writing software for both data analysis and also as an upfront process for how do you design this essays, how do you know where the genome is to target these regions we’re interested in. We have a software product that sits, even before the chemistry is designed or before the primers are designed, and lets us figure out how to focus on particular areas of the genome. It produces reads downstream that are going to be usable in our analysis software.

Carlos: So is that where you come to play? Tell me a little bit about your role at ArcherDX and also we already spoke about this piece of software but is that the only place where technology plays a role or in what other areas; maybe distribution. Is it mainly in the periphery of how the company interacts with others or is there a lot of engineering internally for R&D purposes and such.

Doug: All of the above for sure. As far as my particular role at Archer, I started at Archer in 2013, the company was pretty small at that point. There were two other software developers who had been recently brought on and then me as well so I was sort of like the first commercial full stack developer to come on. The other two guys were both very talented bioinformatics developers and that sort of brought in this other skill set of some genomics background but really understanding like how does a software product evolve and live in the real world when we have a lot of people using it. How do we distribute it, how people are going to use them and so forth. That’s why I was brought on to help launch all the initial software products and then at some point I was “promoted”, in quotes perhaps. Promoted out of development then into full time management that happened probably three years ago. So I helped build up the software team as it exists now. And then most recently I’ve moved into a role where I’m primarily focused on looking at the next big growth phase of our company. I think in true Agile function we develop our software with a very focused need and developed and deployed the things that we needed in that moment. And you know where we’ve been for the last couple of years now and especially now we are looking at really strong dev ops practices. How do we package and deploy our software. How do we get it to a continuous integration pipeline and out to our customers really quickly. And also one of our big challenges is that, as you alluded too earlier, it’s a lot of data that requires lots of computational power to process this data. So going forward how are we going to manage the fairly large amount of computer resources that require to process all of these data.

Carlos: So now, you said at the beginning was two engineers. How many engineers are there now in ArcherDX?

Doug: As far as the software team itself is concerned we’ve got, I think we are around a little more than 20 full time and we have few interns and staff right now, and we have a few racks open so we’re hiring. But I think we’re probably wind up staffing up at probably around let’s say 25ish full time developers this year.

Carlos: The reason I asked that is because I think as I said earlier there is somewhat of fear of this industry. Have you seen that across maybe of some peers that don’t work with you that might be, I don’t want to say fear, that uncertainty do I fit in. Is that something that you’ve seen across maybe others peers that don’t work in the industry?

Doug: Yeah, I think I can sort of directly relate to this as well feeling fairly intimidated coming into this field in which there is a very deep and I guess complex or complicated body of knowledge so I think that’s pretty normal. You know, the message I would convey is that, as long as you are an intelligent person, you have good attitude, and you have solid fundamental software engineering skills, I mean you can basically jump into any domain of knowledge and you could be productive, almost immediately and I was like saying before that can buy you enough time to really get your feet underneath you as far as the particular domain of knowledge is concerned.

Carlos: I think we should do something. We should do a bit of an exercise. Let’s think of the typical commercial application of say, again, a client of ArcherDX. So there is an actual client of mine. We actually work with a couple of molecular genetic labs that do tests for cancer prevention and also for parental planning so forth. I’d like to get an idea of say let’s say for us when we first started working with them. What are we doing in, right? We don’t even understand the big picture of the actual workflow of how a sample came in and how they actually charge insurance and all that stuff. And all the systems in between we’ve been able to now build older systems. We’re very intimate with the process of the life of the patient records. Not necessarily at the lab level that’s where you expertise come in. So give us a bit of a high level of understanding of what happens at the lab because if a anything, a regular lab like our client is going to be the one that buys your consumables. They are the one to buy the sequencers. They are the ones that have the direct contact with the doctors or the actual providers, and the insurance and the patients. They are the ones that will do let’s say the big bulk of this in a commercial way. Of course utilizing companies like ArcherDX or buying products and services from companies like yours. So give us a bit of an idea, so patient goes to the doctor, what happens?

Doug: So patient goes to the doctor and they went in for some reason of course so let’s say they have something like, you know, there is a tumor for example. So what would happen in that particular case most likely is that tumor is going to be biopsied, so they will take a tissue sample of that tumor and where ArcherDX comes in is that if this is a lab that works with ArcherDX they will already have our kits on hand. For example, we sell kits for looking for both DNA variants and RNA variations in solid tumors. So they would take that tissue sample. They would purify that sample so you want to get rid of all the genetic material that isn’t DNA and RNA and then you amplify the regions of that DNA that you’re interested looking at so. Like I was saying you can do what’s called whole genome sequencing and basically what that means is that you’re kind of utilizing the power of the gene sequencer to get reads from all over the genome and then normally what you would with that is assemble that so you could see what a genome looks like which could also sort of use that as, okay we are looking for variations in this particular gene that we know is often involve in cancers like this. You would look at those particular genes and see how many reads cover that particular gene i.e. coming off the gene sequencer. How many of those reads coming off to sequencer actually mapped to the gene you’re interested in. But in whole genome sequencing you just don’t really get enough of coverage. There is not enough of those reads to be confident in a call so that’s why ArcherDX is typically doing targeted sequencing in those cases. So what we do is we have these little things called primers that, they basically stick to single strand of DNA in a particular place and let you amplify the regions that come after so maybe there is hundreds or a few thousands of those different regions. We’ll amplify those basically for the purposes of noise reduction so that that amplify gene sequencer. The gene sequencer is then going to see a whole lot more the things that we’re interested in seeing than it will things that we don’t really care them much about. And then of course in the resulting the called fast queue files but these are the text files that show up at the end of the sequencing process that will contain mostly reads from regions that we are interested in looking at.

So from the patient walking in the door, to getting a data file at the other end of the sequencer is basically that process. And then once we have those data files that’s where the analysis software comes into play. We load those files into our software.

Carlos: Those are the HL7 files, correct?

Doug: Those are different. So what we load into our software are typically fastqueue.gz files. Those are just gzip compressed text files that represent all the reads that came off the sequencer.

Carlos: Got it.

Doug: There are other files as well called bam files that are produced by for example the ion series of machines. Well we can also take those as well. Yeah, so that’s what is going on or into our analysis software and that’s a whole process unto itself. I don’t know for you, I want to dig into that now but I can’t.

Carlos: No, I think I was curious a little bit about the, I’ve heard some issues with HL7. I don’t know if it was related to that.

Doug: Yup. Not in our particular case now.

Carlos: Got it. So let’s say the reads come out of the sequencer, right? But then one of the things that I’m always wondering, are there any patterns that we can recognize from having a lot of sample runs in the past? Like what happens with that data? Is that data somehow reusable in the future to make some sort of analysis?

Doug: Yeah, so you certainly can do things like that. There are different ways to think about this from sort of purely tackling it from the angle of your question, can you use prior analysis to inform future analysis. That’s definitely possible. You know, we call that a meta analysis in this case where you are utilizing data from more than one particular run to inform the results of a particular sample. There are some really specific ways in which we do use that in the ArcherDX analysis pipeline.

Well, taking a step backwards there are errors that crop up in quite a few different places in this whole process. It could happen during sample prep. There’s any number of things that could go wrong and during sequencing as well. You know, sequencers aren’t perfect. I would say error rates are somewhere between like 1 base in every 1,000-10,000 to even as much as like 5% or 10% for certain sequencers and in certain conditions. But basically what that means is that there are errors coming through these data files that if you didn’t do anything to clean those up you would probably call those as a variant in your analysis pipeline.

Carlos: A false positive basically.

Doug: That’s exactly right. Yup. So you know, one way of dealing with that as it applies to the question that you asked is you could basically use additional samples to determine, ok, we know the sequencers produces error, sometimes the sequencer produces systematic error i.e. if you have a whole bunch of the same bases leading up to an error coming from the sequencer. Sometimes the sequences leading up to that can actually produce that error so there is a motif in the DNA that causes the sequencer to misread something, so that can happen. But also it can just be completely stochastic. It may in one read it shows up here and another read it shows up there, and then for the next couple of hundred reads it’s completely clean.

So one way of doing some data clean up is to do we call it outlier detection in our particular pipeline. But basically what it does is it utilizes multiple samples that are run at the same time or it could be samples that have been run prior to this like a normal data set that you would prepare your data to. And what we use that for is looking at how noisy is each base coming off the sequencer typically speaking so we can do statistical analysis of that to say, “Alright, this particular base coming off the sequencer tends to have this much noise or it has this much error rate associated with it.” So if it’s a particularly noisy spot and we know that because we’ve seen it happen from time to time or time and time again. We can set a very high threshold for calling that variant so let’s say 10 of every 50 times we sequence that base it’s an error that means that for us to actually call variant in that position we need a lot more evidence that the variant actually exists. And then conversely there are areas that we would say are very quiet. Where typically there are no errors that show up and so if we see errors that show up there maybe 1 in every 100,000 samples or 10,000 samples. If you see something show up there, the statistical threshold is quite low and you can be fairly confident that even with a low amount of evidence that you should probably pay attention to those results.

Carlos: Are you sure you’re not like a geneticist and biologist and you’re just playing software engineer because you know about at such a depth of the actual science; like I commend you for this.

Doug: Oh, thank you! I think it’s just if you’re excited with what you’re doing and I do love what we are doing, I think it’s just motivation to learn as much as possible about it. There are a million people that know so much more than I do and all of the actual geneticist, biologist at the company know far more than I do but. I don’t know if it would be dangerous and at least talk about that on the podcast.

Carlos: Well, you know, definitely getting that sort of insight from a software engineer and being able to see kind of how you see this world and through your eyes, through the sensitivity of again the developer that as I said earlier you might want to work at Facebook, and re-improving the like button. Nothing against Facebook in doing that by the way but there is definitely a bigger impact when you’re seeing this sort of visceral sort of change in the world. Anyways, this is just something that I admire. I admire what you’re doing. I admire the thing that ArcherDX is doing and all their engineers that are brave enough to jump into this field. This is very exciting stuff.

I think we’re nearing to the end of our episode. I have two last questions for you. What would you recommend any engineers that are interested in the industry that again they might be, even listen to this episode might be a little bit like, “Whoah”, what was all that. How can they learn a bit more to make sure that again if they come and apply at ArcherDX that they have some of their science up to date that they are somewhat writing to this industry. What will be some resources that you might share with us?

Doug: That’s a good question. I guess I would say a couple of things. One, as I have mentioned before, I really do believe and this is little insight into our hiring process here too but you don’t necessarily need to have that really solid domain knowledge. You just need to have a solid technical background but even more important than that is somebody who’s probably having a really good attitude is the number one thing that stands out right off the bat and then to obviously be an intelligent person and a good problem solver. I think if you got those two things and a good technical foundation you are already setting the stage for being successful. You know, obviously being excited about, you know if you are to apply at Archer being excited about what Archer is doing I think is also a pretty critical component because people do the best work when they are working on something that they really care about. But more on a practical level if you are really interested in bioinformatics there are certainly quite a few resources out there. I think one of the challenges with bioinformatics is that it’s just not an understanding of the bioinformatic algorithms like how do genome aligners work, how does the variant color work, things like that. You also need to understand the underlying biology of what those algorithms are actually trying to tell you. So you know, there is a lot of step to dig into but probably if I were to give one particular resource for people to get started with is actually writing bioinformatics code. It think it is at rosalind.info that have the bioinformatics course that sort of starts you from really early on like literally counting the number of Gs and Cs that are in a genetic sequence. But it turns out that that’s actually really important for things like primer design because it affects the melting temperature of those DNA primers. They may seem like they are trivial problems but they are actually not. They are sort of sneakily teaching you about bioinformatics while they work you through this fairly long problem set.

Carlos: I think that if there is people that are suited for this sort of challenge is definitely us, the engineers with that said of course, right? This is obviating the actual scientist. But I think being able to come to this field and provide some of our own powers if you would to support them I think is what’s going to move this field forward, right?

Doug: Definitely.

Carlos: By the way I know I was ending this but I have another question, I am curious. What do you see the future of this? Like right now as you said it sounded kind of futuristic, the story about Zeta unfortunately. But let’s think of like the future like what would be a good future look? How would a positive future look like in terms of let’s say all doctors are using these tools? So yeah, what does the future look like for somebody like everyday person who wants to have a doctor that is using the latest and greatest genetic tools? What it may look like for them if they have again developed cancer or any illness like this?

Doug: Yeah, for sure. I guess this is a capping off of Zeta’s story is that I think there is a pretty good chance that if her cancer had been found much earlier that she might actually be alive today and so I think the lesson there is that if we can pick this things up much earlier don’t give these cancers a chance to replicate themselves into so many companies that they are not really easily treatable we’re going to be much better shape. That’s definitely part of it is trying to transition from this reactive stage to a more proactive stage. You know, as far as future types of analysis in this field, one of them is ctDNA which stands for circulating tumor DNA, this is a great tool for catching certain cancers early or even you suspect some they may have cancer but you can’t even identify where it might be coming from. What winds up happening is that because cancer cells are reproducing themselves faster than normal cells, those cells still die and they decay in your blood stream. If you take a blood sample, you have certain kinds of cancers, you can actually see a signature of those cancers in the bloodstream. It’s a very low rate so you need even more sequencing power to detect these things typically. Maybe you need 10, or 15, or 20 million reads to get enough resolution to see them. But what it means is that maybe in 10 years you’re going to the doctor’s office and they take a blood sample and they run that blood sample through a panel that’s actually looking for cancers as well and they are not just looking at them. A whole bunch of other things I don’t know about. Does this person have a signature of any particular cancer and if you do hopefully there is a therapy to wipe that out in its very early stages such that it never becomes that long term problem.

Carlos: Man, that’s kind of a dream future and I hope it gets there because, I don’t need to speak to kind of talk more about this because it’s a potentially sensitive topic but I hope it get there soon. Again, I thank you for being part of this and spending some time with us on the show kind of explaining it to us. Giving us a bit of an overview as to not only what you’re working on but also potentially how other engineers listening to the show might be interested in stepping into this.

So Doug, thank you so much for joining Tech People show. This was a great episode and can’t wait to hopefully have you back or meet you in person some time.

Doug: That sounds great, Carlos. Either of both of those sounds great to me.

Carlos: Thank you so much.

Doug: Thanks, Carlos.