NewCo Shift Dialogs, in partnership with EY
Massive data problems plus massive markets mean a big opportunity for companies like Color Genomics
Health care is a multi-trillion dollar market awash in data, but thanks to regulatory and political sclerosis, it’s been a difficult sector for NewCos to gain a foothold. But that hasn’t stopped a new wave of startups from trying, and perhaps the most interesting are focusing on the intersection of genomics, lethal disease, and preventative testing. We covered Grail Bio last year, and for this week’s Shift Dialogs, we speak with Elad Gil, a co-founder of Color Genomics, which like Grail is working on new approaches to battling cancer. And as Gil explains in our interview, there’s far more potential for Color’s services than just cancer detection — as genomic testing costs scale toward zero, the potential for saving lives scales up. Below is the interview in both video and text, edited for length and clarity.
John Battelle: Welcome, Elad. To start, what’s Color Genomics founding mythology? How did you and your co-founders come up with this idea?
Elad Gil: Color’s mission is ultimately to democratize access to genetic testing and to do it in responsible way. A lot of the genesis of the company was driven by various personal stories the team have.
We’re four co-founders. My co-founder, Othman, I think has the story that we really collapsed around in terms of why we did this, which is he himself is a BRCA2 carrier, which means he’s at high risk for certain types of cancer. His mother had breast cancer twice. His grandmother died of breast cancer.
Cancer had been something that had been very pervasive in his family. It was due to a hereditary mutation that had been passed down from generation to generation that he himself suffered from.
One of the things that his own personal experience had shown was that even though his mother had cancer and even though she had a strong personal history in terms of lots of family members being affected, testing wasn’t offered to her to understand if it was a hereditary thing until after her second cancer. It took over a year for her to get tested.
It was very expensive to do so. Many of these tests at a time are costing $2,000 to $4,000. It was out of reach for most people. Collectively, we thought that it was really important to think through, “How can you both drive down cost and increase accessibility to genetic test for really important health conditions.”
That was really the origin of the company, and given our backgrounds in software, we ended up taking a very software-driven approach to it.
I want to dig into that, this trend of people who work at very large scale with data and software platforms, as you did at Twitter. Why do you think it is that so many folks who have spent that much time in what is traditionally platform technology, digital technology, are getting into healthcare?
I think it’s three reasons. Number one, there are certain things that you learn at Google or Twitter or other places that I worked or team members have worked that teach you a skill set that honestly just hasn’t been applied yet in healthcare at all. Often, I think the healthcare technology markets are about 10 years behind where everybody else is.
For example, many hospitals are still on their own private clouds instead of setting things up on Amazon because they’re worried about security. But if you think about it, Amazon’s going to be dramatically more secure than anything they’re doing alone.
Number one is you just acquire a skill set that isn’t available at healthcare. Number two, I think a lot of people have very strong personal stories. Everybody’s impacted by cancer, by heart disease, by neurological disorders, Alzheimer’s.
As people get exposed to that, especially as they get older, you get sick family members. You yourself may be affected by something. It causes you to reconsider what you’re working on, but it also means that you start to become familiar with something that you never thought about.
There’s almost a set of entrepreneurial companies that people think about at each stage of their life. In your 20s, it’s all about dating apps and, “Where am I gonna move?” It’s apartment finders and event apps. Whenever you talk to 20-something founders and they’re brainstorming ideas, that’s what they talk about.
As you start to get into your 30s or 40s, you start to worry a lot more about, “How does this impact other aspects of my life?” How often child bearing and other types of services. You see Max Levchin doing things like Glow because of his personal experiences. You see a lot of these types of companies emerging.
There are almost a class of them now in the valley. They started by, like Grail Bio, that’s another, started by people out of companies like Google or Twitter. What are the initial products that Color has come out with?
Color offers a set of services that help you understand your risk of developing hereditary cancer. Then providing information that allows you and your physician to act on that information. Our core component is a genetic test, which covers a 30 genes that most impact your risk of developing the eight most common types of hereditary cancer.
That includes things like colon cancer, breast cancer, pancreatic, skin, etc. We also then provide information that you and your physician can use to craft a personalized health and prevention plan for you as an individual.
For example, if you’re a BRCA1 carrier your lifetime risk of developing breast or ovarian cancer as a woman maybe as high as 80 percent. It means that they’re certain things that you want to do. For example, just getting mammographies at a younger age maybe a step that you should take.
Or possibly something more dramatic. We’ve got the cultural example of Angelina Jolie … It was even higher than 80 percent.
Yeah. There’s different estimates in terms of the impact of these genes. We actually have had very strong personal stories that have happened and it’s actually within our own personal network. For example, one of our investors ended up rolling out Color as a benefit for his company. It turned out that his fiancée was a BRCA carrier. She got tested. She then found out she was positive.
Her mother got tested. Her mother was positive. Her mother actually working with her physician after they have oophorectomy. She had her ovaries removed and it turned out that she had stage III cancer.
In her ovaries?
In her ovaries.
It had not presented in any way that was symptomatic?
It had not presented and they never would have gotten tested, because they fell outside of what would normally have been the criteria for testing.
A great example of that is, if you’re a woman, and you’re diagnosed with breast cancer at the age of 46 instead of the age of 45, then this type of test won’t be covered by many insurance companies. That means you miss it by a year, maybe you started developing it earlier in your 40s. There are these cutoffs that prevent larger scale testing for populations.
The lines were drawn because tests cost $4,000. Insurance companies and medical associations try to be very cautious about that cost burden. As you drop the cost like we have, to something at $250, suddenly that means that you have a much more accessible product.
What is it that’s dropped that cost? Obviously, there seems to be a renaissance of genomic-based testing and various companies doing various things. What’s the core reason that now it’s $200 as opposed to $4,000?
There’s three reasons. Number one is, there’s just this ongoing drop in cost for genomics as an industry. That’s largely driven by this company named Illumina. It’s coming out with cheaper and cheaper technology.
They’re almost like the IBM PC of genome sequencing.
Yeah. Maybe even Intel would be a good example. Every year you get a cheaper, faster, better chip. It’s similar to that. Second is our emphasis on software. We’ve really applied a lot on data science and a lot of workflow tools to dramatically drop the cost that are outside of sequencing. Sequencing is one of your cost components. There’s all the other things that you need to do to do a clinical genetic test.
You have to have what are known as variant scientists, which are PhD level scientists, review every single difference between you and every other person on the planet. Is that disease-causing or not? What are those software tools that you can build to help those people answer those questions, according to what’s known as ACMG guidelines which are national guidelines that have been published around this?
Similarly, when we built a bioinformatics pipeline, we built it the way that Google actually thinks about building its search pipeline, where we have a very modular architecture that allows us to iterate rapidly on both the development and deployment of that pipeline.
Then also validate that it’s working against thousands of samples that we can very quickly say, “Is this still providing the right clinical cause?” It’s a lot of very deep software engineering.
I don’t know whether you’ll reveal this, but what is the scale currently of your operation in terms of how many people have gotten the test? That leads to a second question which is, do you use all of that information to make the whole system better?
We haven’t said much about our scale. I should say that we have very strong proof points around different populations using large clinical studies. There’s a 100,000-women study known as the WISDOM study that’s across the UC system.
There’s others that we’re involved with, as well as we have an employer channel where we’ve been working with other companies and providing Color as a benefit to the entire employee basis.
In terms of using the data, to your point there are all sorts of things that we’ve been doing from a quality perspective that uses machine learning, or other technologies to be able to basically learn off of all the different data that we collect so that each incremental task gets better and better.
Which is exactly how Google solves search problems, and how Twitter deals with timeline issues is, the more data you get the better the acuity for each individual. The trade is that I, as an individual, have to be comfortable giving that data to the common as long as that data is privacy saved. Do you have any regulatory issues with that, and how do you manage that?
We focus on a few things. Number one is we let people control how their data is used. People can opt in or out of use of their data for research or other studies. We follow HIPAA, so we make sure that everything’s secure and privately held.
We also, from a regulatory perspective, have done what’s known as CLIA and CAP. CAP accreditation is the College of American Pathologists and it’s considered sort of the gold standard for laboratory development test. We’ve also done things like our CE mark for European market, and things like that. We’ve actually tried to be very thorough in terms of those tests for regulatory purposes.
It seems we’re at the cusp, and you made this reference earlier about how healthcare is 10 years if not more behind where cutting-edge technology from the Valley is digitally. When it comes to controlling or understanding our own information, having access to it, and using it in new ways, the healthcare system is in its own way. Is that fair to say? Is it overly cautious or appropriately cautious?
It’s a mix of things. There are areas where it’s appropriately cautious. There are certain things you want to absolutely make sure are done right, because you’re dealing with key health information for people. The other side of it is though, I do think there are all sorts of distortions in the healthcare market. They’re really sort of screwed up in different ways.
If you look at the proportion of spend that healthcare is as a part of GDP, it was about five percent in 1960. Today, it’s around 17 to 18 percent of all our GDP in a massively-growing GDP where everything compounding that goes to pay for healthcare. Healthcare is the only industry where new technology drives cost up instead of down. Obviously, there’s something broken.
One component of it is, ultimately, the patient or the consumer is not really the end customer. They don’t pay for it because they’re covered by payers in many cases. They don’t have a lot of say in terms of how things are actually done. There’s a very strong expert on the other side of the table if you’re dealing with a physician.
All those things distort how an individual interacts with the information, even how people think about data ownership. There are all sorts of issues in the market. One example is, people used to view data as a competitive moat in healthcare for genetic information…
Organizations say, “This is now my data. I don’t wanna share it. I don’t wanna allow it to be used to create a larger platform.” That’s starting to change, particularly in genomics, right?
It’s starting to change. In genomics, it’s changing because it’s getting so cheap to retest somebody that that data is no longer a defensible asset. In other words, if it’s pennies to test each person, or dollars or whatever it ends up being, then the fact that you tested a hundred thousand people or a million people it doesn’t matter, because it’s really cheap to do that as well.
Right, that’s very interesting. There must be different uses for what you’re doing beyond the current business that you have as your first model. For genomic testing, there is Ancestry and 23andMe, there’s other companies. They go to market saying, “Find out your Irish heritage,” or, “Find out your ancestors were Vikings.” Is that in your future or are you leaving that for others?
For now, we’re leaving it for others. There’s so much to cover in terms of really important impactful health information that people don’t really have access to today. Our focus as a company is, how do you unlock important preventative health information for people? Your genome is just one data set. There are all these other sources of data in your body that you’re generating.
Right. Your microbiome, your physical activity…
Your microbiome, physical activity, lipid levels, there’s all sorts of things that impact how healthy you are. There’s all sorts of information you can actually act on in a lot of different ways that people don’t have access to. The question is, “How do you unlock all that really core information that’s so important to your health?” That’s our focus as a company.
When you look at a company like 23andMe, didn’t I hear that they got in trouble with the FDA? What is it about your model that means you don’t get in trouble with the FDA, where 23andMe did?
We’re focused on three or four things that are important. Number one is our tests are physician-ordered, and all of our testing for genetic counseling is part of the purchase, which means you always have a healthcare professional in the loop with this important information. That’s number one. Number two is we’ve done extensive validation of our test.
We’ve actually published that data on our website from day one. We’ve been open in terms of showing how we validated it, technologies that we used, how we’ve approached it. Number three, we do something known as next-generation sequencing, which is the new technology which allows you to look in-depth at each thing that you test.
Rather than try and cover a million different things very lightly, we cover a small number of things very deeply so we can make sure that we give you the right result. Obviously every test will have both positives and negatives. It’s part of anything in this area. We try to be extremely diligent in terms of creating a clinical pipeline that allows us to make the right calls and provide answers that can help you form a position.
You said you’re looking at 30 genes right now and you’re focused around cancer. There are a lot more genes and a lot more things. Will your strategy open up a new product line in mental health, or any of these areas that have some genetic determinants?
Ultimately, if I think ahead four or five years, the cost of sequencing your entire genome will be a $100. At that point, that means that you flip from a world where it’s like, “How can I test somebody once for cancer, and once for cardiovascular disease, and once for neurological disorder, and once is for mental health,” into a question of, “I’m gonna do a test once and from that point on, I’m gonna store all the data for you.”
Then, “How do I surface the right information to you, and your family, and your healthcare providers at the right moments in your life that can really impact you with your health?” That’s really I think the future. How do you become that interface by which humanity interacts with this information?
This I’m very interested in, because it has another parallel to our earlier conversation about these technologies and platform. One of the most frustrating battles in that space with Facebook, and Google, and Twitter, is this open versus closed debate. It strikes me that from a consumer’s point of view, what I want is I want my data — whether it’s my lipid, every six months or a year, and it’s my microbiome done every so often, and it’s my genome — I want to be able to flow that through a system and say, “Here’s my information, grok it, and help me out.”
I want to be the gatekeeper of that, but it feels like the system’s built in such a way that there’s multiple gatekeepers all over the place but I’m not in control. Do you think that in five, or 10, or 15 years, the bit might flip?
I don’t know. The reason I don’t know is there’s companies like Color that are trying really hard to move towards that world, where you own your data and you control it. I do think that there’s a lot of players and [owners] of electronic medical records and other parts of the market, where a lot of the other types of data that you mentioned outside of your genome is stored. Those players have a very different type of incumbency than something like we do. In other words, they’ve done these big deals, multi-year deals with large medical institutions, and so for the next 10 years…
They’re in that contract. Sort of like are we ever going to get sports unbundled from the cable channel? [laughs]
Yeah. Good analogy.
You have a new CEO. Tell me about that.
Othman Laraki, who was the co-founder and original president of the company, recently moved to a CEO role. I moved from a CEO role to a chairman role. Basically the big shift has been two things. One is we wanted to clarify decision-making in our organization, because we are co-running the company. We have two other co-founders; Nish Bhat and Taylor Sittler.
We’ve been co-running the company and everybody had to come to both of us for every single decision. Let’s clarify this and simplify it. [laughs] Make it simpler. Secondly, Othman has that very strong personal tie, sort of the patient-advocate viewpoint. He’s a very strong and deep technologist. The thing that really differentiates us and the thing that’s going to shift this whole industry and people’s access to it is software.
Ultimately, how do you think about how software can impinge your own preventative health at massive scale? At Google, where he spent many years — I was there for many years there as well — he was involved with building out the really early developer side of Google Chrome; the browser that’s used very broadly.
He was involved with building out their own internal, what’s known as CDN, basically moving data around the world, across the Internet. He has that very strong technical background that we think is really important for the future of the company.
You have quite a star-studded list of investors. At some point or another they’re going to want a return. How can you scale what you’re currently offering to the point where you see a tip, and be a profitable enterprise that returns to the investors?
We all think that there’s a huge opportunity here, because fundamentally we are in the very early innings. It’s like the first 10 minutes of the game kind of thing in genetic testing. Great example, BRCA 1 and 2, which are the best-known genes, after 20 years of testing, only about 10 to 20 percent of people who are carriers for those genes know about it.
80 percent of the most tested genes in the US, or amongst the most recessive genes in the US, have no idea…
You really are in the first pitch of the first inning.
We’re very early on. It’s like starting Amazon. I don’t remember if it was ’94 or ’96, or ’95. Or somewhere around there. There are these early signs that this is really a massive shift that’s happening, in the way that people can think about precision medicine, preventative health, tailoring things to themselves based on their genetics. We think it’s going to be all pervasive because those costs keep dropping.
One of my favorite all-time studies was when McKinsey was hired by AT&T to study the cell phone market. They told the AT&T — this was back in 1988, and they tried to extrapolate out the market and had these big brick phones, and they cost thousands of dollars. They said the total market size for the US for cellular was going to be 800,000 people.
So AT&T didn’t invest in it.
Somehow McKinsey keeps getting those contracts.
Yeah.10 years later, or that sort of time frame, [AT&T] ends up buying McCaw Wireless for 20 billion dollars, because McCaw built out a mobile network. I do think we’re at a similar moment where there’s this huge shift happening. We’re going to capitalize on that, but also I think software is the way to do it because that’s the way that you can truly get to mass scale.
Now the curve ball question which we ask everyone. Which is, if you could travel back in time with the knowledge you have now, to when you were 20 years old and give yourself some advice, what would you tell yourself?
I guess I would tell myself to go start a company. What I did is I went and got my PhD, and then I joined a few startups, and then I joined Google and eventually I started my first company that was acquired by Twitter.
In hindsight I think that there’s a unique period in your life where you really don’t, in many cases, have a very high burden in terms of financial burn. You have a lot of free time. [laughs] That’s actually the best time to start something. I wish I just started something earlier.
Might have been a dating site. [laughs]
Yeah. Or apartment apps, who knows?