Non-Human Genomics: Interview with Zal Bilimoria

I recently caught up with Zal Bilimoria (@zalzally) of Refactor Capital, a seed-stage venture capital firm that he and David Lee launched in 2016 to attack “fundamental human problems” in regulated fields like health, agriculture, finance, aerospace, etc.

Despite his software background in building monetization and mobile products at Google, LinkedIn, and Netflix, Zal has taken an aggressive interest in healthcare and biotech products since helping launch the $200M Bio Fund at Andreessen Horowitz. We dove into his current investment thesis on genomics, which emphasizes the vast opportunity for applying genomics beyond healthcare to agriculture, consumer products, industrial processes, and more.

Read below to learn his framework for understanding the sector and the opportunities he’s most excited about…

EKP: Zal, you’re particularly excited about the impact of non-human applications of genomics. Why?

ZB: Look, we’re just at the tip of the iceberg when it comes to breakthroughs in genomics and the businesses that will spring from them. I can’t emphasize that enough. The twenty-first century is going to be the era of genomics and biology.

At the technology level, there are 3 drivers of this:

  1. Moore’s Law — the cost of sequencing a genome has dropped and continues to drop, exponentially. From $1M in 2006 to ~$1,000 in 2016, for example.
  2. Machine learning — I hate using buzzwords, but yes “machine learning” enables us to analyze the genomic data much more efficiently than ever before.
  3. Gene editing — we’ve made it possible to accurately add/remove individual genes and at relatively low costs. Primarily through CRISPR. The advancements made by Jim Collins at MIT, Jennifer Doudna at Berkeley and their teams truly are the stuff of legends. The Cas9a enzyme was just the first breakthrough with CRISPR. An even newer discovery this year — CRISPR-Cas13a — targets RNA with an order of magnitude greater precision than Cas9, which targets DNA.

Primarily, the excitement in mainstream media and Silicon Valley is about the “human applications” (aka healthcare) of genomics. When we see CRISPR on the cover of Newsweek and the Economist, it’s part of this controversial debate about designing your child to be taller or maybe having blue eyes instead of brown eyes, for example.

So, why am I excited about the non-human application of genomics? Well, we’re years away from modifying human genes, and nearly every innovation in the realm of medicine requires extensive regulation that create barriers to adoption. It’s transformative and we’ve invested there heavily, but it’s going to take a bit more time.

What’s possible right now and advancing at a faster pace are the non-human applications of genomics, since they face far fewer regulatory hurdles and/or take advantage of regulatory seams (which we love at Refactor).

As a society, we’re beginning a shift from chemistry to biology in creating almost everything around us. Most physical products we interact with and consume are comprised of organic materials that we’ll be able to edit at the gene level, and many of the synthetic chemicals we use in industrial and consumer products will be replaced by safer, organic compounds we design and manufacture biologically. Think of the polyurethane in the foam of a surfboard or the chair or couch you’re sitting on, or the moisture-wicking fabrics you might be wearing.

Give us a mental framework for understanding the types of new companies here.

First, let me explain the difference between sequencing and genotyping, as it’s important to understand and not transpose the two (like I did when I first got started 4 years ago in genomics).

A good analogy here might be the Netflix mobile apps since I used to work on that product. At the top of the Netflix app, you have 2 ways to find a movie or TV show: you can Browse or you can Search.

  • When you’re browsing, you’re in discovery mode — you’re hoping to find something interesting, but you don’t know what it will be. We’re sequencing entire genomes or large segments of it in order to find new insights in the data such as which genes correlate to which conditions, diseases, or traits, otherwise known as characterization. In humans, we’re quite early in this process, having characterized less than 5% of all known 30,000+ conditions and diseases (nevermind the many 1,000’s of traits). This is why advances in next-generation sequencing (NGS) have been and will continue to be instrumental to the advancement of science.
  • When you’re searching, you know exactly what you’re looking for. This is analogous to genotyping, when you already understand which genes correlate to specific conditions and your aim is to identify and quantify it. Our ability to modify genes means we now have the power to invent or fundamentally alter the inputs, outputs, and/or processes used in the development of nearly any product made with chemicals, but using biology.

Now that we’ve explained genotyping vs. sequencing, let’s discuss two categories for how genomics is putting its stamp on non-human applications: “genomics driving diagnostic development” vs. “genomics driving product development”.

Genomics driving diagnostic development

Sequencing is often done in research or academic environments because it’s about answering questions (or even figuring out what questions to ask) rather than providing new solutions. But, there are many commercial applications for collecting and analyzing this data, and the possibilities only increase as the cost of sequencing drops precipitously.

These companies help us better understand the world around us. One of the companies we invested in here is Trace Genomics — think 23andme but for soil. They sequence the microbiome of the soil and identify the beneficial micro-organisms and pathogens that constitute a particular soil sample, then understand how those microbes interact with various seeds, fertilizers, climates, etc. to get the best yield on a crop. The company’s goal is to create the most comprehensive soil health database on the planet. Disease is multifactorial. That’s the value of the database — to discover organisms that can ultimately be characterized to certain conditions, diseases, and traits.

Another Refactor company, TL Biolabs, created a faster, cheaper, and more comprehensive diagnostic for livestock. They’re helping ranchers understand for every calf born on their farm today, what their milk yield is likely to be, the protein and fat composition of their milk, and who their parents are (since they often don’t know who the dad is). Traditionally, this has been very expensive, and thus cost prohibitive for a rancher to get diagnostics on every calf. TL Biolabs has reduced that cost dramatically with their novel genotyping technology.

Did you know the average cow is more sequenced than the average human?

We’ve spent a couple decades sequencing livestock because the universe of conditions, diseases, and traits are relatively narrow in terms of what we need to know to create the best dairy and beef cattle. Since we know which genes correlate to milk yield, for example, we can also quantify the amounts of milk, fat, and protein that a particular cow could produce during its lifetime. That’s why TL Biolabs is revolutionizing genotyping technology and driving the cost so far down that every calf born on a farm can now be genotyped with this company’s solution.

What this all comes down to is providing vastly more data to make more informed decisions. Diagnostics lead to treatments and/or prescriptive decisions.

Genomics driving product development

The other realm of activity is where teams are using genomics to create new products (think “atoms” rather than “bits”): these are physical products sold to consumers or enterprises…materials, fuels, probiotics, pharmaceuticals, personal care products, and industrial chemicals like hydrogen peroxide.

One of our Refactor portfolio companies, Solugen is using genomics to create an ultra-safe, ultra-pure version of hydrogen peroxide, one of the most commonly used compounds in all consumer products — it’s likely in every disinfectant and cleaning product in your home, and it’s used in higher concentrations and/or purity levels for bleaching paper, rocket fuel, and cleaning semiconductor fabrication facilities. Hydrogen peroxide is a $4B market, and what Solugen has done is create an ultra-pure version of H2O2 using only air, water, and plant starch. Solugen’s proprietary enzymes are produced using CRISPR-Cas9, which accelerates the production of this critical compound within their patented bioreactor.

It sounds like there’s the same B2B vs. B2C split in business model we see in software. Can you explain those 2 approaches for genomics startups?

Sure. Within the product development realm of genomics, startups can take two paths:

  1. Create full stack products for consumers, like Soylent, a beverage that can be used as a snack or a meal for consumers that’s powered by genomics. (B2C)
  2. Sell the enzymes or the bioreactor technology to scale the production of enzymes, which allows you to be a platform for other enterprises to create their own products. Full stack products can also be manufactured and sold to enterprises. (B2B)

Some startups want to do both (B2C and B2B), and that’s a really tough position to put yourself in because you’re dividing your efforts and resources; the lack of focus could be a recipe for disaster. Deciding which is the smarter path for a given business depends on the team’s expertise, their technology, the target industry, and so much more.

As an example, Soylent could have gone in the platform/B2B direction, but their founding team knew how to bring a product like that to market and how to craft a brand that excites consumers. The full-stack strategy was the right move for them. And now that they’ve established themselves as the top product in their category, they could — if they determined it was the right move, of course — license their IP to other companies and also create a fantastic business doing that, likely with immediate interest from the top 20 food producers of the world.

This covers a vast scope of industries. Is there one market opportunity you find most compelling from a business standpoint and/or an impact standpoint?

David and I started Refactor to hunt for companies with both huge market opportunity and huge potential for impact, and from that perspective, one area I’m especially excited about in genomics is the opportunity to eliminate the use of petrochemicals in everyday consumer products. These are massive markets — billion-dollar opportunities for founders.

Not to promote our own portfolio companies too much, but what Solugen is doing with hydrogen peroxide is the perfect case study…

From a straightforward safety perspective, understand that there are roughly 100 facilities around the world that produce hydrogen peroxide — each costing $100M to operate over a 5–10 year period — and one of them blows up every year. EVERY YEAR. The unstable nature of hydrogen peroxide, especially at higher temperature or concentration levels, makes it high risk, even in stationary storage. For example, when power was out during Hurricane Harvey in Houston and plant temperatures rose to dangerous levels, it caused an explosion. With respect to high concentration H2O2, the US military, for example, is one of the biggest purchasers of hydrogen peroxide, using it at 85–98% concentration levels for rocket and torpedo fuel.

In the bigger picture here, one of the main inputs for creating hydrogen peroxide as it’s done today is petroleum. We’re talking many, many millions of gallons of petroleum and petroleum-based products used annually to create hydrogen peroxide. Not only is petroleum a scarce resource with price volatility, but it’s, of course, a fossil fuel contributing to carbon emissions and other environmental damage…and we’re using petrochemicals in thousands of everyday consumer products.

Solugen is acquiring corn starch and using CRISPR-Cas9 to offer a new form of hydrogen peroxide (Bioperoxide) that’s safer and purer. They subsequently launched their first consumer “full-stack” product — a brand of household cleaning wipes called Ode to Clean that’s nearly sold out its first batch of inventory.

“Genomics” is a buzzword in Silicon Valley. There’s a lot of hype, mainly from people without a background in biology. How do you filter out the BS?

I wouldn’t call it identifying BS — it’s more about scientists in a lab who’ve found something interesting, and they see potential commercial applications for it. But, they raise money for what’s really just another R&D phase with years of further experimentation still ahead. That’s the scenario I find myself filtering out among genomics startups. Which ones are really ready to create a product and a company over the next year? We don’t have much interest in funding experiments, we’re interested in funding new companies who are ready for commercialization and have a clear path to revenue by their Series A.

How much of the market opportunity will be captured by startups versus large incumbent companies who spend billions on R&D?

I think there’s a lot of room for new companies to become the category winners. In machine learning and artificial intelligence, there are these incumbents — Google, Facebook, Amazon — that have such a distinct built-in advantage based on the amount and sophistication of data they already have to work with in creating general software products; sure, there might be an opportunity for startups but much of the board will be won by the big guys.

In looking at opportunities presented by genomics — whether with diagnostics or new product development — I see incumbent companies who have built platforms of chemical products like DuPont, BASF, 3M, and Monsanto not having the data advantage or the wherewithal, or being stuck in an “innovator’s dilemma” where driving these innovations could cannibalize their existing revenue streams. They’re just not wrapping their heads around this “chemistry to biology” revolution that’s underway; it’s not the direction they have long planned to progress in.

The corporates are also having a harder time recruiting the top talent in this space. Because the costs dropped so dramatically, the path of founding a genomics company — or joining a young startup — is much more compelling to researchers at Stanford and elsewhere. You don’t need to work with the big budgets of DuPont or 3M to work on the biggest, most challenging innovations…in fact, you’ll likely make faster progress on a small, independent team.

Does IP create defensibility for startups? Is that a greater factor in your investment analysis than it would be for software startups?

I think it can create a short-term moat in some cases but not long-term defensibility. It takes a big pocketbook and years of litigation to successfully defend your IP against infringement, so that’s not an option for a startup even if they have a legitimate case to make. I do see startups trying to fly under the radar for longer in genomics though — avoiding much publicity until after the Series B when they have a foothold in the market and enough funding to defend themselves if a major company challenges their patents.

Real defensibility derives from the “idea maze” that the founders navigate. This is a concept that technologist Balaji Srinivasan wrote a paper on. He was the co-founder/CTO at Counsyl and spent time at Andreessen Horowitz. Imagine that you as a startup founder are entering this maze and you keep hunting for the way out, bumping into lots of walls, then suddenly see the exit and sprint out of it at a completely different (read: faster and more efficient) velocity. The process you went through testing all different routes and becoming familiar with the maze now gives you a distinct advantage over someone else entering it for their first time. You’ve gained an instinct for how the market is evolving and what is/isn’t feasible that enables you to avoid wasting time pursuing less effective strategies, which is usually what newcomers and corporates end up doing.

Who are the peers you respect among tech investors also pursuing this thesis on non-human genomics applications?

Vijay Pande, who I worked with at Andreessen Horowitz in setting up the Bio Fund. Vijay is a professor of chemistry, biology, and computer science plus a two-time startup founder, and so, much of the mental frameworks I have in this space stem from my time with him. Of course, IndieBio and Y Combinator are both fielding compelling teams at the accelerator level…they’re selecting some really interesting companies and teams. A few other VC firms at the forefront here have been GV, particularly Andy Wheeler.

Aside from regulation, are there specific barriers you’re concerned could hold back innovation here?

Well in food-related applications, genomics suffers from a marketing problem. “GMO” is a negative term among consumers, but it shouldn’t be…everything we consume has been genetically modified going back decades based on simple decisions of what to produce more and less of. And that’s been beneficial for humanity. That we can make more advanced genetic changes to crops and livestock nowadays is great for human health at both an individual and population level. I liked the manifesto Soylent put out proclaiming how their products are “Proudly Made with GMOs”.

The point being, we need a new term that casts these innovations in a positive light. Perhaps “Powered by Genomics” or “Genomics Inside” (as a hat tip to Intel’s very successful “Intel Inside” marketing campaign) or something catchier. Open to ideas!