Deniz Kural, Seven Bridges CEO: Startups are the way to make big ideas come true

BY ON March 16


Series A investment of $45m put Seven Bridges under the spotlight in the last few weeks.

Besides securing the funds to support the development of Seven Bridges platform for biomedical data analysis, the investment brought on board one of the most influential figures of Chinese IT sector, Kai Fu Li, as well as the former Senator Tom Daschle.

During my visit to Boston, I had a chance to speak to CEO of Seven Bridges, Deniz Kural, who generously offered more than an hour to answer the questions I had about the business, biotech, and the future of crucial research done in the battle against cancer.

Deniz is originally from Turkey. The story of his friendship with Igor Bogićević, co-founder and CTO, started about fifteen years ago on the world wide web. They hadn’t met in person before starting  the company. Igor tells us in his way:

“Deniz and I met approximately fifteen years ago at IRC channel, one of the underground networks. During those days we used to code together a lot. , and We stayed in touch all those years, even when our paths parted.

Deniz enrolled in the undergraduate Mathematics program at Harvard, and I decided to lean toward software engineering.

During his PhD program, we started discussing the fundamental premises that would  later become our product – the price of sequencing will keep dropping in the upcoming years and someone will need to solve the issue of storing and analysing the data. Only in this way could medicine and science use the wealth of data being collected.”

In 2010 Igor and Deniz officially expanded their friendship into a business partnership. They met in person for the first time that year at a bus station in Pula – after ten years of an online friendship.

How did you find yourself in the role of the CEO of successful startup?

Accidentally. I was a scientist and I never really wanted to start a company. I never had any interest in entrepreneurship – there were prominent entrepreneurs in my class at Harvard, Zuckerberg for example.

I never had an interest in building a company from scratch. I just wanted to scale genomics.

I never started a company before and in hindsight I was very naive, as I didn’t know all the challenges and problems we’d encounter. If I knew them, I wouldn’t have started.

Never had I managed people before and it’s very hard having a large team. I didn’t understand the scale – and amount of effort that goes into creating a large team, operations and organization. That was the biggest challenge for a young person who had never had a job before.

You graduated with a degree in Mathematics at Harvard – what drew you closer to genetics and bioinformatics?

My interest was encouraged during the early 2000s, by professors George Church and Martin Nowak — both of them globally recognized scientists. As soon as I realized that sequencing was the way to dig deeper into diseases such as AIDS and cancer, I signed up for Church’s class, and later joined his lab.

He invented various types of sequencing, including multiplex sequencing. I went to his lab to build a sequencer called the Polonator, the second generation sequencer used for the Personal Genome Project. I was only a junior, in a large lab, so I was learning from everyone.

This was in the early days, 2006 and 2007, when then the world’s next generation sequencers are being launched, so noone has proven human genome sequencing instrument yet.

George licenced many of his patents to other companies, but the instrument used to be limited.

I joined the 1000 Genome Project in 2008. At the time, this was the most extensive catalog of genetic variability of humans.  I realized that the amount of data these next-gen sequencing machines were producing were going to be incredibly large. Initially the problem was that we couldn’t get the human genome sequenced – now that we could, the data wasn’t going to fit anywhere.  and once we could, it didn’t fit anywhere.

What enabled DNA sequencing to progress?

In 2008, people realized that we could get 100s ofterabytes of data with these machines. That wasn’t possible before then.

Computer power became good enough and it was finally sufficient to do image gathering.

High resolution CCD cameras became more affordable at the same time. There was a drastic drop in price. A lot of the sequencing companies buy their cameras from Hamamatsu, a photonics company,, and they cost about 20-30,000 dollars – that’s  cheap. These instruments were once only used  on satellites.

Fluorescent technology and fluorescent chemistry improved dramatically in the last 10-20 years.

There were chemistry breakthroughs, technology-related microfluidics, fluorescent optical imaging processing power related. Everything became possible around 2008.

The thing that’s overlooked is that timing and luck play an important role in a startups’ success. If you’re too early you fail. If you’re too late you fail.

After graduating from the PhD program at Boston College, you didn’t choose a path in academia. Why is this?

I wanted to build a very large software system for analyzing genomics. I knew that I didn’t want to wait until I got grants for research and that it would take 50 years until I became a professor. So I started the company.

After 1000 Genome Project, it became clear to me that Universities, with their prevailing computing facilities and teams, were not going to be able to really handle the scale of the analysis.

You needed a hundred or 200 computers to do the analysis, and it would be inefficient for each lab to invest in so many since  they would only receive occasional use. Most places don’t have much funding, so it was a perfect problem for the cloud.

That’s what the cloud is designed for – rent a 100 computers or a 1000 computers for 4 days.

That’s hard to do for a lab. Even if they went to the cloud, it’s a significant bioinformatics and IT challenge. Most labs don’t have necessary bioinformatics expertise to know which tools to use.

You use more than 20 different tools going from raw sequencer output to the list of variants. If the lab wanted to do it, they would have to download, compile, figure out the right parameters, and optimize the pipeline. It’s a lot of work and they’d need to hire more people. It’s a giant budget issue.

It seems to me that you believe startups are a better way to make big ideas come true than academia is.

I think so. If you want to make sure that your ideas are implemented and become a global force, starting a company is the only way to do it.

This is because you could have the best idea and others might not pick up on it and build it. Companies like SpaceX show this. Elon Musk could write a paper and say people should build a new rocket.

Doing research is not enough – you need to go and make it happen. I had a lot of research ideas and I wanted to see them happen in the world.

What was the greatest challenge starting up?

Our biggest challenge in the early days was getting a critical mass to build and launch our first product. If we had a prototype product in the market, we knew people would see the value and we’d attract more interest.

It felt like a race against time. We didn’t get paid anything or were on low salaries. I didn’t take any pay until two years after we founded the company.

What about your first customers?

Getting our first customers was very easy for us, but that was also a false signal.  We literally tweeted one day, and got our first customer the next day. It’s a small field and it was a novel product – people had never seen anything like it before so there was early interest from the genomics industry.

Then, for the whole year, we had only ten labs using it and everyone payed very small amounts. That’s a very different segment than enterprise. Our first enterprise sale came much later.

I would like to say that enterprise is very complicated. It’s not that your product needs more features, it’s that the enterprise culture is conservative – they want to know that they don’t spend their time on a company that’s just going to go away.  They first want to see who’d be the winner.

It’s easy to make decisions in a startup – you just get your buddies together and do stuff. Enterprise is slow – at a time they like your product; if they decide to pilot it, this will take a few months. Purchase orders, procurement, security… the whole process takes six to nine months and has nothing to do with your product.

I think this leads to people getting prematurely discouraged. All these steps are logical, but there’s a reason why a lot of people don’t appreciate them. If your business is hard to build and hard  to be taken away, it’s a good business to be in.

Those who follow the U.S. presidential campaign could see that some of them mention promises and opportunities of biotech and genetics. How do you look at the relationship between industry and politics?

Government is the only force that can encourage everyone to share information.

This is why President Obama announced the launch of the Cancer Moonshot initiative, and it’s the reason why UK is doing the 100,000 Genomes Project.

We need these databases and precision medicine cohorts, so that when a new sample comes in, we can analyse it against everyone else and personalise it based on what we learn.

We want to provide a system where one patient’s cancer could be compared against 50 million other cancer genomes. Using statistical learning, machine learning and previous cases, the doctor would make recommendations for treatment – these are exabytes of data.

Even Bill Clinton said during one speech that 83% of all cancers will be cured in 20 years.

It’s very realistic. Cancer is a thousand different diseases. There’s 600 genes involved in various processes and pathways influencing cancer. Out of any of these 600 genes, any given cancer can have 2-15, or even 2-50, of these genes mutating.

It’s trillions of combinations. Each cancer is unique. I always recoil a little when I hear the phrase “cure cancer,” because it’s really a thousand different diseases and the way to cure it is to cure one at a time. The progress will also look like that. We’ll cure one at a time and when we cure 800 different, totally different diseases, we’ll cure cancer.

Some cancers are already pretty much cured, such as CML. We can cure it because it’s a simple cancer. It’s caused, most of the time, by a single mutation. If it’s a complicated cancer, there are 30 different ways it can kill you.

What are the greatest challenges Seven Bridges is facing at this stage?

I think one big challenge is finding  ways to share the data that benefits people while at the same time  preventing abuse of data sharing. Some groups are too conservative – they don’t share patient data. But without that data, we cannot help the patients.

There’s no easy answer, but it’s a challenge that we s and other companies will continually face.

We don’t define ourselves as a software company –we’re a science company and we care about the quality of analysis. It’s incredibly important to respect privacy rights and patient rights. We’re proud to be able to work with governments on two different continents. That shows that we’re able to meet both scientific standards and the standards expected in genome data sharing.

Interview originally published here.

For fresh updates regarding SEE tech startups follow us on Twitter.

Add a comment