NOTE: This story originally ran in the June 13th issue of Bio-IT World, viewable here.
By Kevin Davies, Bio-IT World
June 13, 2012 | ALISO VIEJO, CA – What makes Ambry Genetics different? It might be the relaxed surfing culture at the genetic diagnostics firm’s southern California headquarters, the humorous candor of founder and CEO Charles Dunlop, or the firm’s early commitment to clinical exome sequencing -- even if it means cannibalizing its thriving diagnostics business.
Last November, Ambry Genetics debuted its CLIA Exome sequencing service at the International Congress of Human Genetics in Montreal. The initial price was $14,500 for a trio of samples (for example, a patient and both parents). Dunlop has quietly dropped the price to $7,900, which is clearly below breakeven.
“If the reputation of your company is for raking every penny out of every rag, then you’re not going to be used,” says Dunlop. His attitude is this: “If we can cannibalize our own markets, then let’s do it. We’re applied scientists. Let’s not lose the goal. We don’t have a bunch of guys with financial calculators sitting over our heads, so let’s take advantage of that.”
In addition to the original Clinical Diagnostics Exome, the company has also launched the First Tier Exome – a first pass at a subset of some 4,000 known Mendelian disease genes in the proband only -- for $5,000.
“But we’re not just CLIA exomes,” says Ardy Arianpour, Ambry’s vice president of business development, who exudes the self-confidence and zeal of an investment banker (see Box 1). “I like to say, we got the iPad, the iPhone and we got the iMac,” says Arianpour.
“Sometimes we need to follow Ardy around with an oxygen tent,” says Dunlop.
After listening to the story of Ambry’s early years, it is remarkable the company is even in existence, let alone forging a growing business in genomic medicine.
Dunlop enrolled at University of California San Diego to major in physics, but says his student years were an annoyance. “My father cut both [my brother James and I] from the money,” says Dunlop. “He drops us off with $1,000 and says ‘Bye.’ We had to scramble, so I got a job flipping bagels… I was that jerk at college going from couch to couch, always at parties. I had nowhere to go!”
To survive, Dunlop got a job at the Scripps Research Institute in Gregor Sutcliffe’s lab. (He figured Sutcliffe, the first person to sequence pBR322, would be a good mentor because his was the most impressive office on the top floor of the institute.) Dunlop quickly showed his aptitude for lab work, so Sutcliffe doubled his salary. After six years of nonstop work, Dunlop started his own clinical diagnostics company.
In 1999, Dunlop persuaded his brother James to leave Lockheed Martin and join him in the business. Dunlop began by focusing on cystic fibrosis (CF) testing, primarily using gene mutation panels. “We thought, why don’t we launch a sequencing service and see if we can get some samples.”
James had the clever idea of contacting the CF Foundation and offering one free sample per institution. The Foundation circulated an email announcing that a CF lab was offering limited free testing for indigent patients. “I swear to God, two days later, there was a box of samples!” says Dunlop. “Oh my God! We’ve got a business!”
The first sample to arrive belonged to a newborn from UCSF. The Dunlop brothers were dumbstruck. “What do we do with it?” Dunlop recalls saying. “We all took pictures. It was still in the diagnostics specimen bag.”
Nevertheless, the company came close to wiping out on several occasions. Suspending their own paychecks, the brothers would often frequent the happy hour at the local Mexican restaurant, order a beer and eat all the free enchiladas on the menu. “Isn’t that ridiculous? It was that bad,” says Dunlop, laughing now.
Struggling to make payroll, the Dunlops resorted to desperate measures. “We’d go out at night and find the after party,” Dunlop recalls. “Everyone in Laguna is wealthy, and James could get us in anywhere we want.” Dunlop scouted the prospects while his brother handed them the business plan.
On one dire occasion in 2001, with vendors angrily demanding payment, James stuffed a dozen business plans into a backpack and headed into town. He returned the next day with a business-saving check for $15,000 from one sympathetic family.
After growing the CF testing business, Dunlop realized he could expand the service for pancreatitis by adding a couple of additional genes. Ambry has grown organically into new markets since then, and currently offers some 300 tests on a variety of platforms – Sanger sequencing, next-gen sequencing, microarrays, even some good old-fashioned RFLP testing. “It’s a combination of low tech and high tech, frankly. We just want to solve the problem the best way,” says Dunlop. He proudly notes that staff counselors will steer potential clients to a more rudimentary test if that’s appropriate, putting the client first.
In keeping with Dunlop’s surfing obsession, Ambry has launched a non-profit foundation – the Mauli Ola Foundation -- to raise awareness about genetic disease and teach CF patients how to surf in the process. (Salt water aids lung function in CF patients.) “We rent a bus, and invite the best surfers in the world to take these CF kids surfing,” says Dunlop proudly.
Dunlop surfed semi-professionally in his younger years, and several childhood surfing buddies are on the payroll. He sprinkles surfing vernacular liberally in conversation, like the time he was competing against one of the sport’s big names. “I saw him hyper time and went Ka-pa-pa-Pow! I’ve never seen someone hit the lip like that. My friend said, ‘Oh my God, you’re going to lose this heat.’ I paddled back out, decided to do my thing. You know what? I won that heat because I caught three waves and he only caught two.”
There’s a moral to the story, Dunlop tells his more than 200 employees: We’re going to surf our heat, guys. Don’t worry about the echoes from other groups. “Let’s do things on our pace and launch [tests] when they’re ready. That was a huge turning point. It calmed everybody down. It was a leadership moment. And it made the [clinical] exome possible.”
“It’s easy to launch a sequencing test and validate it but the customer service and reporting the physicians and patients need is another matter. We’ve been good at innovating. I couldn’t give you an overall business model; we struggle to get it done that way. But we have a really good reputation in the clinical world.”
Dunlop takes pride in that Ambry never holds formal management meetings. “I hate meetings,” he says. “They’re a waste of time and I’d rather make a few mistakes here and there and run through walls.” A major strategic decision is more likely to occur during a chance encounter outside a bathroom than in the impressive, underused conference room.
Enter the Exome
Arianpour says Ambry cut its teeth for 2.5 years running research exome sequencing projects, thanks in part to an early deal with Agilent for its SureSelect target enrichment technology. “We got on board, blew it up on research, and at ASHG last year launched the CLIA Exome.”
“We were turning people down if they said they had patient samples,” says Dunlop. “We said, ‘You have no idea what you’re getting into.’ I wasn’t willing to sacrifice quality -- let’s do this correctly.”
Leading the charge on the bioinformatics pipeline was Xiang Li, who burned the midnight oil for the better part of nine months (see Box 2). “We were confident this was going to work -- but how fast it was going to work internally was the question. When we found out it was that easy, I said, ‘Screw it, let’s halve the price. Let’s get the patients what they need,’” says Dunlop.
Dunlop is satisfied with the early progress, although there was plenty of anxiety around the first result. Some of the first patients were sequenced in collaboration with the Kennedy Krieger Institute at Johns Hopkins. “We’re ending diagnostic odysseys,” says Dunlop.
He recalls seeing Li walking nervously into his office. “I swear, I came out of the bathroom, he’s running up the stairs saying, ‘We’ve nailed it, we’ve nailed it!’ I ran into Xiang’s office. There were three mutations, and it was just obvious! We did co-segregation to confirm. How rad is that?!”
One of the biggest headaches is variants of unknown significance (VUS). “They’re great in research but a nightmare when you’ve got a sick patient in front of you,” says Dunlop. “I’ll match any group in the world,” he says. “We know what we’re doing in sequencing. When everyone else gets their exomes, they’re having teams of scientists going through the data, then geneticists review it. Our bioinformatics staff is narrowing the [the process] down to a couple of months.”
Dunlop suggests that Ambry is seeing higher success rates than the 5-20% reported by other groups, but a key factor is the selection and quality of the sample as much as the sample prep and informatics pipeline.
Ambry tries to consult with physicians in an ethical manner if it feels that a particular patient might not be a strong candidate for exome sequencing. “I don’t like getting negative results. I like it when there’s a clear family history and members for testing,” he says. In some cases, health insurance companies are covering the CLIA Exome service, says Dunlop, though it is a delicate process.
When it comes to the competition, “Me and Ardy can lose sleep, but no-one else,” says Dunlop. “I want the people here to be focused on solving the problems they know to solve… Let’s just help these patients. And maybe make a little money along the way. It’s that simple. We’re a medical diagnostics lab – we’re just helping people. Why obsess about business models? We don’t want to be living paycheck to paycheck anymore. It’s that flippin’ simple.”
The CLIA Exome service is but one of several launches in the coming year. “It’s more exciting from a science perspective than a business perspective,” says Dunlop. “The genome is obviously the Holy Grail of experiments… That’s the dream when it’s economically ready.” From the business perspective, however, there are other things Dunlop is more excited about.
The name ‘Ambry’ comes from the family dog. (Dunlop’s first choice, ‘Genomic Health,’ was already taken.) Dunlop recalls his sister, studying at Berkeley, leaving a 7/11 store with a pack of cigarettes. “A guy in a van with dogs in a box asks for a cigarette. She says, ‘I’ll give you the whole pack if you give me one of those dogs.’ He reaches in the box and says, ‘Here, take Ambry.’ So literally some gypsy named my company Ambry!”
James offers that “’Ambry’ is a place where you find solace or peace of mind in a church or cathedral,” which seems fitting somehow
Ambry will move to whole-genome sequencing when the economics dictate, but Dunlop is in no rush. “How much more advantage do you get?” he asks. “Right now, we’re doing the simplest thing possible on the genome – finding the one disease-causing, co-segregating mutation,” says Dunlop. “To really do the dream, it’s one patient -- all data known. But we’re really far from that.”
BOX 1: Clear Path to CLIA Exomes
Ardy Arianpour joined Ambry Genetics in early 2009 as employee #19, following stints with Clinical Data and Cogenics. The company’s rapid growth has been strategic, with alliances with sequencing companies including Illumina, Ion Torrent, Raindance Technologies and Roche.
“It is truly amazing to look back and see that we have grown to 200+ personnel in such a short timeframe and all organically with no investors. The good ole American way of how business should be done,” states Arianpour.
“It’s the wave of the future,” says Arianpour of CLIA exomes. “Now we can do a clinical diagnostic exome. We try to do it ethically -- trios wherever possible.”
Ambry’s traditional diagnostics business consists of more than 300 tests, with particular strengths in cancer and cystic fibrosis as well rare genetic disorders including Marfans syndrome. “There are a lot of diseases where there isn’t much support. We give that support.” 20 months after the launch of research exome sequencing, Ambry launched clinically. It began with a sequencing panel of 81 X-linked genes for mental retardation in early 2010, developed on the RainDance system.
Ambry seems content to specialize in exomes rather than rush into whole-genome sequencing. “We leave that [market] up to Illumina, Complete Genomics and BGI,” says Arianpour. “We don’t compete in that market. WGS keeps getting cheaper, we can’t compete from a reagent standard. It’s not our core business… Everyone’s trying to do whole-genome sequencing, but sometimes less is more (powerful).”
Arianpour says that Ambry is working with several “highly prestigious institutions and hospitals that call us.” There have also been a number of international requests, which surprised company management. “Some countries take this even more seriously than the U.S.,” says Arianpour. “All the technologies and companies developing this stuff are in the U.S., but there’s not availability outside the U.S. We’ve got lots of international requests.”
If the informatics department reports a suspicious mutation, Ambry staff will take a deeper look in an R&D setting. “Until we’re 100% certain, we’ll report and give the information to the patients. Having the expertise in house sets us apart from anyone else doing something like this,” claims Arianpour.
BOX 2: AVA the Variant Analyzer
From the clinical standpoint, Ambry decided to write its own interpretation software and tailor it for its own needs. “Anyone can do sequencing. Our value is on the informatics side,” says Arianpour.
The Ambry bioinformatics R&D team is directed by Xiang (‘Sean’) Li, a molecular biologist by training who obtained his PhD in bioinformatics from the University of Illinois before joining Ambry two years ago.
Much of Li’s time last year was spent developing the crucial in-house genome interpretation algorithms and pipelines. “We had many non-stop nights,” says Li. “Three months work non-stop.” The result: a homegrown clinical analysis tool, AVA (Ambry Variant Analyzer), which is applied to each variant identified within the diagnostic exome.
Li’s team decided to build its own genome analysis pipeline – from making variant calls and annotating the variants to filtering the variants and distinguishing causative mutations from the haystack of some 100,000 raw variants. He credits the seminal papers in 2009-2010 on the discovery of Mendelian mutations from the groups of Jay Shendure and others on Kabuki Syndrome and Miller Syndrome.
“There was no available software that could fit our needs,” says Li. “We tried all of them, but none of them was perfect for us. [We see] Lot of indels, but they [other programs] can’t reliably detect them.”
Through an iterative development process, Li and colleagues steadily improved their in-house software. “We de-bug, we fix it. We leverage internal software, find the problem, took them next cycle to finish,” he says. Upon receiving feedback from Ambry clinical geneticists, Li’s team would make adjustments and fix various bugs within hours if not minutes. Reported mutations undergo interpretation by board-certified medical directors that review and sign off all reports.
In the menu of algorithms, Li says Ambry makes use of predictive in silico programs like SIFT and PolyPhen, but critically, he says, is “we don’t use them for filtering. That’s a key difference. They may cause some false negatives.” From a large pool of candidate DNA variants, Li says, “if we filter them out by mistake, then we fail the project… Sometimes I’m surprised people use those programs to do filtering.”
In addition to using dbSNP and 1000 Genomes data, Li says Ambry has the advantage of having hundreds of internal genome samples serving as normal controls. “We have a big pool of control genomes – we have done next-gen sequencing for years, so we have a lot of samples to compare,” he says.
In a recent success story, Ambry studied a family with retinitis pigmentosa, a very heterogeneous form of hereditary blindness in which more than 100 genes can be mutated. After years of fruitless research, Li and colleagues identified the causative mutation in a couple of months – an indel caused by a large 48 basepair deletion/7 basepair insertion. “It was our first case! Very comforting,” says Li.
Arianpour says that Ambry is receiving numerous exome samples each day and the volume is increasing. “The jump is much larger from single gene analysis to exome sequencing than the jump from exome to whole genome,” says Li. “Whether to go to whole genome really depends on cost and infrastructure and throughput.”