Every morning, Josh Chase makes his way from Goetz House, a yellow-and-white cottage in the middle of nowhere Washington, past a makeshift 8-foot “security fence” built from square-cut pieces of lumber and black plastic tarp, down to the greenhouses. A sign hangs on the fence: “PERSONS UNDER 21 NOT ALLOWED ON THESE PREMISES.”
He’s on his way to tend his buds. At 25, Chase is the second oldest of four recent college grads who co-founded Amerifarms, a marijuana startup headquartered in Washington’s “Kush Valley.”
The team has four greenhouses stuffed with 2,000 plants — 54 varieties in total — and a proprietary nutrient cocktail that’s supposed to bring out their fattest and most potent nugs. “We’re weeding through to see what [plants] work best with our system,” said Chase, a former financial analyst turned Amerifarms’ master grower. “It’s not just about selling exactly what the consumer wants. As a grower, we have to develop what’s new. We have to figure out the next new things.”
In the Bay Area and Washington state, for example, Girl Scout Cookies — a top-shelf hybrid
variety of pot known for its minty, skunky aroma — is really popular. The young company is trying to predict what the next generation of Girl Scout Cookies will be.
That process requires a good and stable growing environment, some intuition about how consumers’ tastes will shift, and, perhaps most importantly, a solid understanding of the plants’ genetics. Amerifarms says it’s got the first two. The industry is still trying to figure out the third.
This is the beginning of the Green Rush. Genetics is poised to help entrepreneurs create better, more powerful, and sometimes personalized strains, which can be classified into a Yelp-like rating system that’s actually rooted in science. But which entrepreneurs? There’s a battle brewing between open-source stoners and Big Weed, and who wins might determine if the pot industry is led by a company like Monsanto or one like Tesla.
The Grateful Dead, ChemDawg DNA and the Amazon cloud
In 2011, right as the legalization movement was picking up steam in several states, Kevin McKernan left biotech giant Life Technologies, the company that had acquihired him just a few years before. On his way out, he signed a non-compete promising he wouldn’t tinker with any DNA-related projects.
“We couldn’t pick up a pipette,” McKernan recalls, “except for cannabis because they couldn’t touch
that.” He founded
Medicinal Genomics, a company focused on marijuana genetics.
Still, he needed to be cautious. The feds weren’t as tolerant as states like Delaware or Connecticut, which had just passed medical cannabis legalization and marijuana decriminalization laws, respectively. His own state, Massachusetts wouldn’t pass medical marijuana laws until the following year. The lawyers told him to keep his paws off pot, in the U.S. anyway.
So McKernan, a veteran of the Human Genome Project, slipped a DNA purification kit through airport security and jerry-rigged a lab in the Dylan Hotel in pot-friendly Amsterdam. His mission was simple: take a sample of a cannabis plant, macerate it, isolate its DNA, and take that back to the States for sequencing. He’d done DNA extraction many times before, though never in a hotel room. Still, he succeeded. He traveled back with a container that housed Chemdawg’s DNA. (He didn’t declare it.) DNA is just information, so it’s perfectly legal to work with, even if it comes from cannabis.
McKernan, a Grateful Dead fan, had chosen Chemdawg because story had it that this potent pot strain could
trace its roots back to a 1991 Grateful Dead show during which one DeadHead paid another $500 for some good weed. The pot came with the seeds that would later sprout Chemdawg, though its genetic lineage was still a mystery 20 years later. McKernan posited its “folklore could be cemented [through] genetics,” he said. If all went smoothly, Chemdawg wouldn’t be the only plant to emerge from genetic limbo. Its genome could give McKernan clues about the origins of other strains, and, down the line, maybe even help settle the debate over whether indicas and sativas were one species or two. It would be the first time the plant’s genome was sequenced.
Back in the U.S., he ran the DNA slush through a sequencing machine that spit out hundreds of “reads” — or jumbled snippets — of the Chemdawg genome. Less than three months after leaving Life Technologies, McKernan had the beginnings of a genetic map for cannabis.
Unfortunately, those short chunks of DNA had to be assembled into the full genome, and, at the time, that wasn’t so simple.
The plant’s genetic code is rich in adenine and thymine — two of the four DNA building blocks. Together, they make up roughly 65 percent of cannabis’ nucleotides. In humans, each of the four nucleotides is represented almost evenly. That means when you get cannabis reads back from a sequencing machine, it’s very hard to put them in order, even with the help of computers. Everything looks the same.
“If the jigsaw puzzle all looks blue,” McKernan says, “then your algorithm can’t solve it any better than [humans] can…It became a real mess to try and assemble. We couldn’t get the genome sequence any better. The technology couldn’t make it come together.”
In essence, the valuable information hidden in Chemdawg’s genome was still beyond reach.
He put the data up on the Amazon cloud for anyone to download, hoping someone else might have a successful go at it. Amidst federal government crackdowns on growers, McKernan decided to lay off the pot research, and instead shifted his attention back to sequencing people’s genomes.
From a stoner’s curiosity to science specimen
A few months after McKernan unloaded his cannabis DNA reads on the web, Jonathan Page, a plant biologist at the University of Saskatchewan in Canada, published the
first paper with a draft sequence of the cannabis genome.
Page had been working on cannabis since the ‘90s. His early work focused on the chemical pathways that produce marijuana’s two best-known chemicals: the cannabinoids tetrahydrocannabinol, THC, and cannabidiol, CBD. In the aughts, he started doing some small-scale studies of individual genes in trichomes, the hairy, pungent parts of the female marijuana plant that produce THC. Then in 2009, while he was in San Diego on sabbatical, he got an email from Tim Hughes, another Canadian researcher interested in cannabis.
In the email, Hughes mentioned that whole genome sequencing was getting cheaper and that it might be time to tackle weed’s genes. He asked whether Page — or people he knew — would be interested in collaborating on a cannabis genome project.
“In those days, there wasn’t a lot of research into cannabis. We were a voice in the wilderness,” recalls Page, who’s now the chief science officer of a marijuana-focused biotech startup. “The plant was a curiosity for a lot of people — for stoners and others — but there was no industry per se, in the sense we see today.”
In those days, there wasn’t a lot of research into cannabis. We were a voice in the wilderness.
- Jonathan Page
It would take Page and Hughes’ team almost two years to publish their findings on Purple Kush, another popular strain of marijuana. They encountered similar issues as McKernan, but they were able to stitch its genome together. Then, they compared that to hemp, pot’s non-THC-producing cousin. Among other things, they found that the enzyme that produces THC was present in hemp’s genome, but inactive. They posited that this genetic variation stemmed from hundreds of years of selective breeding for the psychoactive compound. Without knowing the underlying genetics, growers had altered the plant’s genes significantly, not unlike what agriculturists had done to livestock and other cash crops.
“It was almost like a crowd-sourced breeding program,” said Page. “All throughout the world, people were looking for unique and potent types of marijuana, crossing them to produce better and better strains adapted to the conditions they were growing in. It’s sort of an amazing success story in terms of plant breeding.”
Page and Hughes’ work was a first step toward understanding the history of that breeding program and building cannabis a family tree. Their paper made headlines on pot publications and mainstream sites like
LiveScience and
Nature. Together with McKernan’s work, the study laid the groundwork for the pot industry’s nascent genetic revolution. Suddenly, the idea that breeders, growers, scientists, and companies could peer into the plant’s genome and figure out how to adjust it to create new designer strains more quickly than ever before seemed like a real possibility.
More than four years on, the industry is still trying figure out how to cash in on that promise. The financial incentive is there. The legal marijuana market is worth about $1.5 billion, according to a
2013 report by the ArcView Group. It was expected to grow to roughly $2.6 billion last year and to more than $10 billion in the next five years. That’s despite sweeping drops in the price of weed.
The financial worth of its genetic information, says Leslie Bocskor, a managing partner at Electrum Partners, a consulting firm focused on the pot industry, will be “in the hundreds of millions of dollars without a doubt.”
The holdup is that the science is still embryonic due to federal restrictions on research.
As individual states push for legalization, marijuana remains on the
U.S. Drug Enforcement Administration’s list of schedule 1 substances, sandwiched between LSD and ecstasy. Despite some research in animal models suggesting cannabis may have therapeutic effects, the federal government still considers it a drug “with no currently accepted medical use and a high potential for abuse.” That makes it almost impossible to study for anyone not researching addiction, even if all they want to do is study its genes.
“It’s the most widely used illicit drug in the world, but it’s hard to get permission to work with it, and that continues to this day,” said Page. “There’s this huge science gap…This is a plant that, because it’s been left behind by the research community, in some ways [we’re] starting from scratch.”
Weeding out weed
At the Amerifarms farm, Chase, the company’s master grower, is in charge of editing their stock of 54 varieties down to roughly 20. To do that, he and his team will nurse plants until they produce flowers that can be tested in a lab for THC, CBD, and terpenoids, a class of compounds that give each strain its signature flavor and scent. Apart from these chemical profiles, Chase is also looking for plants that grow fast, have high yields, and are relatively short so that they’re easier to manage inside Amerifarms’ greenhouses. Based on all these data, he’ll pick out which strains he wants to keep, and which ones get the boot.
The stakes are high. This isn’t a science experiment. The company, which has a permit to grow marijuana commercially for recreational use from the state of Washington, took out a hefty loan to jumpstart the operation.
Selecting the wrong plants will impact how well they do financially, and that selection process is slow. The team has to wait three months for plants to bloom flowers, which make the stuff people want to smoke. Only then will they know whether they’ve got a winner or a dud. The technology is no better than what agronomists were using decades ago in other industries.
For pot, this is pretty standard. It’s also inefficient — another byproduct of the lack of research into weed as a plant.
But as cannabis emerges from the black market, these inefficiencies won’t do. They mean that Amerifarms — and other marijuana operations across the country — are wasting money, electricity, water, and other resources on products that may be junk from go. (This, of course, has other environmental repercussions.)
Being able to figure things out earlier — when the first leaf sprouts or from the seeds themselves, would be a tremendous boon to the industry. In the future, the growers with the biotechnical chops to produce the best plants most efficiently will have the advantage. Right now, it’s all about a race for knowledge.
“It’s still the moment before the race starts. Everybody gets to jockey for position,” Bocskor, the pot-focused investor, told me. “Large businesses that would be interested — big pharma, tobacco, alcohol — are currently staying on the sidelines. As soon as that changes, the entire market will change. Right now, the playing field favors the nimble and the agile. Later, agility won’t be as important as inertia and heft.”
It’s still the moment before the race starts. Everybody gets to jockey for position. Right now, the playing field favors the nimble and the agile. Later, agility won’t be as important as inertia and heft.
- Leslie Bocskor
The road to precision pot farming
So, new startups are cropping that promise to help growers optimize their yields using data, smart algorithms, and science — the types of things that have changed other industries in recent years. They’re hoping to establish a customer base they can grow before what many see as the inevitable countrywide legalization of pot — and the industry’s subsequent takeover by big business.
For example, PotBotics, a Palo Alto-based company, is testing out gene radars, a Star Trek tricorder type of technology developed by scientists at MIT.(He declined to name the researchers because the company signed a non-disclosure agreement with them.) Using image recognition linked to characterized genetic traits, the tool can analyze snapshots of seeds and infer some of their DNA makeup — and how their genes might manifest themselves as plants with different characteristics given different growing conditions, claims David Goldstein, one of the founders.
He says PotBotics’ scientists have already tested it on tomatoes, peppers, eggplants and corn, with good results. Now, they’re developing the pot version, which is dubbed NanoPot. They’re setting up growing facilities where they’ll A/B test different light sources, soil types, UV filters, and irrigation methods affect the yield and chemical production of various strains. They’ll take that data, pair it with their images and genetic analyses of seeds, and use software to come up with optimal growth plans for each strain.
“Most growing operations hire a master grower with [pot-growing] know-how,” says Goldstein. “It’s all based on intuition, so it’s hard to duplicate the results of what they were able to achieve.” The NanoPot platform would eliminate the guesswork and variability.
Intuition honed through years of experience, some say, is what makes high-end marijuana special — an artisanal product that isn’t mass produced. But that romantic notion doesn’t sit well with some.
How do you know what you’re getting? There are guides online, but it’s impossible to know for sure. Image: Kent Hernandez/Adrian Saravia.
In explaining the problem, you often get a comparison to the wine and alcohol industry. When you buy a bottle of Merlot or Port, you know with certainty what grapes and regions these wines come from — likewise with a bottle of tequila or bourbon. Yes, they might differ a bit from year to year, but if you’re a wine connoisseur, there are ways to judge that too. The vintage is printed on the bottle.
That doesn’t exist right now in the cannabis market. Several small-scale studies have shown that the same strains of cannabis bought from different dispensaries in the U.S. can have wildly different amounts of cannabinoids. This is true too for identical strains bought from the same vendor at different times. Whether those variations are due to growing conditions or strain counterfeiting isn’t entirely clear. Genetic testing — and labeling — could help with that.
Currently, some dispensaries do display THC and CBD content, so it’s possible for consumers to track that, but these are just two of the more than 80 cannabinoids the plants produce, some of which are thought to augment the effects of THC and CBD. That means the labels are really incomplete.
So, what you have is chaos, a term one of the scientists I interviewed used.
In the medicinal marijuana market, especially, that’s a big deal. Patients seeking a consistent experience from one toke to the next have a hard time getting it. Harlequin, for instance, is often recommended to adults and children with epilepsy because it’s supposed to be low in THC, but high in CBD. (CBD doesn’t get you high and is thought to help reduce seizures. Charlotte’s Web is another popular strain high in CBD, but since it’s doled out as an oil, it’s easier to control its chemical makeup.) Jessica Tonani, a cannabis entrepreneur,
recently found that roughly 1 in 5 Harlequin samples actually tested high for THC.
Verda Bio, Tonani’s Washington-based startup, is working on a DNA database to help growers create varieties of pot that will consistently produce certain amounts of cannabinoids. Page’s Anandia Labs in Canada is creating a similar product that’ll inform growers on how they can use selective breeding to tweak the plant’s natural genetic variation. The end goal is to create strains optimized for specific medical conditions, like anxiety, cancer or pain. (He and others are quick to say they’re not going after genetically modified plants — plants into which new genes are inserted, for instance.)
When they’re ready, the key will be pairing genetics to standardized growing conditions, like the ones Amerifarms and others are engineering in their greenhouses.
Inside one of the Amerifarms greenhouses.
“The genes are a very important determinant of a plant’s cannabinoid profile, but they’re modified by the environment,” Page, the geneticist who sequenced Purple Kush, said. “Doctors want to know that the material that was consumed by their patients this year was the same as last year’s. That’s the challenge, and it can be achieved by stringent control of the environment in indoor growing. If someone asked me what the future of cannabis growing will be — it will look like that.”
Most of the precision-farming technologies people told me about are still in the test phase, and to make things even more complicated, human genetics also matter. “Each person is likely to experience the same strain differently based on the human variation in their cannabinoid receptors,” McKernan, whose company is trying to pair human and cannabis genetics to recommend strains to patients, told me. So seeking information from review sites like Leafly and Wikileaf amounts to “reading tea leaves,” he added.
Ethan Russo, a former president of the International Cannabinoid Research Society and former medical advisor for GW Pharmaceuticals, a company developing cannabis-based medications, put it more bluntly: “We are dealing with the wild west of cannabis therapeutics. There is some real snakeoil out there.”
Crowdsourced Pot
When evolutionary biologist Daniela Vergara finished her doctoral thesis in 2013, she started hunting for a new project to take on as a postdoc. She had the option of studying sunflower genetics with Nolan Kane, a plant biologist at the University of Colorado, Boulder. He was using sunflowers to study domestication and adaptation to extreme environments. Vergara had been spitballing the idea of working on cannabis since 2009, and now the chance to move to Colorado, where recreational marijuana would be legal come 2014, seemed like the perfect opportunity. She convinced Kane to let her study weed.
“‘With all the changes in the law, we could answer some of the same questions in cannabis’,” he recalls her saying. “She felt that was more novel. I think she’s right.”
That discussion set the stage for the University of Colorado’s
Cannabis Genomic Research Initiative (CGRI), a loose association of a few researchers and citizen scientists aiming to finish what McKernan and Page started: decoding the cannabis genome.
Even in pot-friendly Colorado, doing research turned out to be complicated. The scientists applied — and got — a permit to work with hemp that produces low levels of THC, and they’re just starting to set up a greenhouse. They also wanted to analyze the THC-rich strains you’d find in a dispensary.
But the legality of handling THC-producing plants for research purposes is murky. So, the scientists had to get creative, or else they risked being shut down. Vergara and Kane’s team taught breeders to extract DNA from plant matter. Some breeders have even built their own mini-labs to help out. The breeders would then hand them the DNA for sequencing.
In about a year, the CGRI has built a crowdsourced library of hundreds of DNA samples from a wide variety of strains. So far, they have sequenced 24 genomes and are hoping to finish 16 more in coming months. Vergara has begun to compare these to cannabis’ closest relative, hops — othe ingredient that gives beer its flavor — and to examine the genetics of the plants’ mitochondria and chloroplasts.
These tiny cellular structures have their own mini-genomes, and they can sometimes tell scientists a lot about a plant’s evolutionary history. But in the case of marijuana, they look pretty similar from strain to strain, so as far as lineage goes, they might be pretty useless. But Vergara thinks that hidden in the mitochondria and chloroplast’s genes could be clues as to how sex is determined in cannabis. In the world of pot sex, it’s not a simple X and Y chromosome scenario. It’s not clear how a plant ends up female or male. Vergara’s hypothesis is that there may be some interplay between these micro-genomes and the rest of the plants DNA that could help determine whether a plant will be male, female or both.
“In the future, if we find out that there is a correlation between sex and these types of [genomes], we could tell people whether there’s this likelihood that your plant is going to be a female or male,” she said. That would be extremely valuable because males in a grow operation can throw things off considerably. Growers want female plants that produce salable flowers. If a male somehow infiltrates the grow room after females have started to bloom, they could pollinate those flowers and start the seed-making process. The yield would be shot. Sexing plants is a very big deal in the pot industry.
The scientists have also set up a website called
CannaData where people can submit pictures of cannabis leaves from around the world, along with information on the strain, sex, height and the growing environment. The idea is to build a public database that links leaf shape to other morphological data, and ultimately genetics. This will all help scientists and growers better predict a plant’s chemical profile with more accuracy in the future.
The hard — and expensive part — will be analyzing all these data. The researchers can’t accept money from cannabis-related businesses because technically they’re breaking federal law, and public funds are largely unavailable. Here too, the researchers have had to turn to the crowd. “It’s been pretty neat how much passion there is,” said Kane.
The project, so far, has been funded almost entirely by private donations totaling about $10,000, and a small army of volunteers is helping to build custom open-source analysis software and to analyze the data. (Kane and Vergara’s salaries are paid by university funding for new faculty.)
“A cool thing about cannabis is that everyone has opinions about cannabis — good or bad,” said Vergara. “We hope that if the public is engaged in our research, they can understand why we’re doing it, what it takes and how science works.”
The cannabis family tree
In Portland, twelve hundred miles northwest of the GCRI in Boulder, geneticist Mowgli Holmes, the founder of biotech startup Phylos Bioscience, is embarking on a similar project, dubbed the
Cannabis Evolution Project. Instead of relying on citizen scientists to do the extractions — which could compromise the quality of the DNA — Holmes is partnering with professional labs, who then ship him genetic material. Phylos Bioscience is fronting the bill for now.
Last week, he and his team finished sequencing their first 1,000 strains. They got their hands on DNA just like Vergara and Kane — through donations from growers and collectors with preserved samples, some thousands of years old.
Holmes and his team are using these 1,000 genomes to build a free online interactive family tree for cannabis that people can use to see learn about the history of their favorite strains, plus a freely available database.
Holmes ultimately wants to use his cannabis family tree as the basis for a commercial test to certify the identity more than 10,000 strains of weed already on the market — and any new ones that come along. He imagines a ‘genetically certified’ stamp attached to every strain. The stamp would be displayed in dispensaries and give buyers that sense of certainty that the cannabis industry is sorely lacking. McKernan’s company, Medicinal Genomics, is building a s
imilar genetic identity test.
Holmes is working with labs to get his test, which he estimates will cost roughly $50, added to the list of quality-control tests growers have to get in states where recreational or medical marijuana are legal. That’s how the company will make its profits.
Holmes hopes that by making the data public, other scientists can do more cannabis research. He also wants to prevent the corporate takeover and patenting of cannabis. Holmes is sharing his database with the Open Cannabis Project, a group that’ll be using it to keep track of all the strains currently in the public domain. The database “will serve as public record of already existing strains that makes it impossible to patent them,” he wrote in an email. The goal is “to stop companies who might try to patent them in the future. If you ask me, that’s a pretty valuable use of genetic data.”
But that may or may not be the case. “If the varieties have been used or sold for more than a year, they can’t be patented anyway, irrespective of whether there’s a published database of cannabis genomes,” Mark David Janis, the director of the Center for Intellectual Property Research at Indiana University’s Maurer School of Law. “It sounds a bit lavish to claim that the database itself will curb the patenting of existing varieties. I don’t think that…would preclude the patenting of new cannabis varieties, although the database could certainly serve as evidence of what is already known.”
Back in Kush Valley
After months of working, Amerifarms’ Joshua Chase and his team have weeded out about 5 strains since they started their recreational marijuana business in September. He wants to whittle that down to the best 20. For now, most of these will be made into oils.
But soon, the young company will start selling their flowers, the part of the plant people smoke. Each strain will be sold individually and marketed for its special properties.
“We have genetics that other people don’t,” Chase told me. Right now, we have to take his word for it. But if the work the scientists and entrepreneurs I spoke to pans out, you’ll be able to know for sure.