How I learned to edit DNA in less than a day


In the movie GATTACA, Ethan Hawke’s parents visit a doctor’s office to choose the embryo of their second child. They have a choice between four babies—two boys, two girls—each of which has been engineered to perfection. As the doctor says: “[These embryos have] no predispositions to any major inheritable diseases. All that remains is to select the most compatible candidate.”

Gattaca_Visit_To_Doctor_OffidceScreenshot from GATTACA

GATTACA was produced in 1997. Although we can’t yet choose our child’s intelligence by modifying its embryo, gene editing has come a long way since. We have sequenced the entire human genome. We have discovered how to write biological material like software.

I expect biological engineering to be one of the most important technological advances in my lifetime, so I wanted to learn how it works in practice.

Fortunately for me, New York City has a community biolab called Genspace. Community biolabs are labs at which ordinary citizens can learn about and work on biotechnology. After taking an introductory course in biotechnology in November 2017, I signed up for a course in gene editing this February. The technique we would use is called CRISPR-Cas9.

What is CRISPR–Cas9?

CRISPR-Cas9 is a gene editing technology inspired by nature. CRISPRs—short strips of genetic code—are part of a defense method that bacteria have developed to protect themselves from intruding viruses. When a virus intrudes into a bacterial cell, a protein inside the bacterial cell can identify the virus (using CRISPRs) and cut it in two, disabling the virus.

In 2012, scientists at UC Berkeley first successfully modified this method to target a custom, programmed gene sequence. In 5 years since, CRISPR has proliferated, so much so that you can buy do-it-yourself CRISPR kits online.

The first two minutes of this video by MIT explain the mechanism well.

What did I do in my course?

The goal of the course was to modify yeast’s DNA. Yeast naturally looks white; we had to color it red and make it green fluorescent under ultraviolet light.

This metamorphosis consisted of two steps. First, we had to disable the gene that made the yeast look white (if this gene was not expressed, the yeast would look red instead). Second, we had to add new DNA to the yeast that would make it green fluorescent. CRISPR-Cas9 can do both—disable existing genes by cutting them and inserting new genes by adding new nucleotides.

To do this, I must have been a biology expert, right? Far from it. I took 3 years of high school bio and never did serious lab work. I did not take any biology after the age of 14—I could barely remember that animal cells have a nucleus and bacterial cells don’t. Now I was going to modify the DNA of a living organism? Yes.

So how would I turn our standard white yeast into a red-and-green, disco-loving creature?

White to Red

White_to_RedOur goal was to turn yeast from its natural white color (left) to red (right). Photo courtesy of Will Shindel, Genspace’s instructor.

We had to cut the gene that turns the yeast white. The cutting of DNA is done by the Cas9 protein. This protein must be instructed to go after the right target—in this case, the yeast’s gene that gives it its white color. To instruct the Cas9 protein (think of this as the scissors) we had to add a piece of guide RNA. Guide RNA is like a barcode: it tells the Cas9 protein what sequence of DNA to look for, and once it finds it, the Cas9 protein will make the cut. So, in addition to the Cas9 protein (the scissors), we had to add guide RNA (the barcode).

Minutes 1:50 to 2:50 of the same MIT video explain this process of cutting existing genes using Cas9 proteins (scissors) tailored with guide RNA (barcodes).

Using CRISPR-Cas9, you can now modify this process to go after any genetic code you like by modifying the barcode.

Amazon meets Biotech

CRISPR_directScreenshot of CRISPRdirect, a website that helps you to identify the guide RNA sequence with the highest likelihood of cutting your gene successfully.

How do you create the barcode? This part is fascinating. You simply search on Google for the gene of the organism you want to modify—in our case, the gene in yeast that gives it its white color. (The gene is called ADE2). You then download the gene’s DNA code—a series of letters A, T, C, and G—and paste it into Snapgene, a simple program that helps you read the DNA more easily. You then past the DNA code into CRISPRdirect, to identify the guide RNA (the barcode) with the highest likelihood of cutting the DNA properly, and then order the guide RNA (the barcode) using a website like IDT. At that instant, machines will start whizzing hundreds of miles away to synthesize the genetic code that you just submitted through a form on a website. The RNA is delivered to your doorstep within 24 hours.


Introducing the scissors and the barcode was enough to disable the yeast’s red color. But changing white to red was only half our goal.

Special Effects, Please

The second step was to make the yeast glow bright green. To do this, we had to add extra genetic materials.

After the Cas9 protein (scissors) cut the yeast’s DNA at the site instructed by the guide RNA (barcode), the yeast’s DNA two strands (double helix, remember?) will naturally try to heal. This is when you can introduce new genetic material into the DNA.

The trick is to hide the new, foreign DNA in other genetic materials that the yeast recognizes as naturally occurring. It’s a bit like hiding your dog’s medicine inside his food, hoping that he will eat the medicine unknowingly by finishing his food.

We wanted to add new material to code for our green fluorescent gene. This material was a series of 1510 nucleotides. (Nucleotides are single molecules that humans codify by the letters A, C, G, or T, which stands for the molecules Adenine, Cytosine, Guanine, and Thymine.) To both ends of the new gene that would lead to green fluorescence, we added DNA that matched both sides of the cut in yeast’s DNA. Normally, after a cut, the two strands of yeast DNA will naturally heal back in their own place, as explained in the previous video. However, when enough “repair material” is present in the yeast’s cell, the yeast’s DNA will heal instead by connecting with this new repair material. As a consequence, you have now successfully inserted a little bit of new DNA into the yeast’s original code.

Repair_TemplateThe blue, double strand is the original DNA; the purple, single strand of DNA is the inserted genetic material. This screenshot shows how the purple, new DNA is being connected with the outermost 4 base pairs onto the blue, original DNA.

In my mind, I compare this mechanism to how magnets work. The naturally recognized pieces of DNA that you add to both sides of the newly introduced gene (green fluorescent, in our case) are like magnets. These magnets were chosen to have a strong attraction to both sides of the cut of the original DNA. When the green fluorescent gene with magnets on both sides comes close to the original DNA cut by CRISPR-Cas9, the magnets on the side of the green fluorescent gene connect with both sides of the original DNA’s cut, and as a result, the updated yeast’s DNA now has a line of code that makes it green fluorescent.

Minutes 2:50 onward of the same MIT video explain the process of inserting new genetic materials into existing genes well.


A week after we added the cutting sequence (to cut the yeast’s ADE2 gene that made it look white) and the DNA that would make the yeast green fluorescent to the yeast, we returned to the lab to look at our results.

Our yeast colonies had replicated, and most samples showed red yeast instead of white yeast. Unfortunately, none of the colonies turned green fluorescent under UV light (despite what the image below seems to show—if the green fluorescent gene would have been adopted, the yeast colonies would have shown much greener).

Petri_Dishes_CRISPRPetri dishes with yeast colonies under UV light. The petri dish top-center and bottom-right show mold growth (this is contamination).

A possible reason why the green fluorescent gene was not integrated was that the ADE2 genes were indeed broken (hence the shift from white to red), but that they reconnected with a different sequence, and that therefore the green fluorescent protein (GFP) was not adopted.

Accessible Science

Now that I have edited live genes, what are my reflections?

It’s hard to believe that somebody with no deep background in biology can understand and learn how to edit DNA in less than a day. I don’t suggest that I have mastered the discipline—far from it, of course—but I have learned how to practice the basics.

The technologies used to modify DNA are relatively simple too. The tools used are relatively simple—either pipettes, to dose liquids, or devices that spin, heat, or cool the genetic material.

Will We Edit Human Embryos Soon?  

Chinese researchers have started to edit human embryos. And last summer, the United States followed suit, with the first American editing of a human embryo (using CRISPR-Cas9), see the video below. Like in GATTACA, will all babies soon be edited?


In the short term, I don’t think so. Our understanding of the human genome is too limited. We know how some genes code for some features, but we are far away from knowing what genes make you smart, tall, or strong. 23 and me, a service that synthesizes your DNA for $199, will tell you if your urine will smell after you eat asparagus, or if you’re likely to grow bald, but it will not tell you your IQ or whether you’re good at public speaking.

Our understanding of human genes is expanding rapidly though. Almost every issue of New Scientist reports on the discovery of a new gene. Stephen Hsu, VP of Research at Michigan State University, recently led a study that can predict the height a person will be within a 3 centimeter range based on the person’s DNA.

So as we learn more about genes, will we allow “editing” of humans?

I think the answer is yes. If you are the parent of a child that will be born with a terrible, hereditary disease, and it is possible to save your child from that suffering, would you not?

Even if some countries don’t allow it, others will. Wealthy people who care about giving their children the best possible genes will go to the countries that allow for gene editing and use the technique to modify their embryos.

Initially this will be done only to disable hereditary diseases. But what about modifying genes to upgrade ourselves: making our children more intelligent, better-looking, or stronger?

That will happen too. Practices will spring up in which the ultra-wealthy can “upgrade” their embryos. Some people will choose to do this, because the surest way to feel you’re leaving something for future generations is to improve the chances of your offspring being successful.

If you’re in that position, what would you do?



Location-based philosophy

Location-based philosophy

Since moving to Colorado, I spend more time outdoors than before. On the flipside, I spend less time meeting new people, and less time making ideas happen. Why? Because living in the mountains influences what you think of and what you’re invited to.

Where you live defines how you live. This realization made me think back of the places I’ve previously lived. I realize that the definition of a good life varies for each location.

So what does it mean to live a Good Life?

Aspen, Colorado: a good life is … being outdoors

A perfect week for a Coloradan includes a climbing-adventure, rafting down a river, and backpacking—ideally all together with friends. Most men in Colorado grow beards, so I stand out perfectly as a beardless European. Work, for many, is second in priority for most to the life outdoors.

New York City: a good life is … chasing a dream

When you stand still on a NYC-street, observing the crowd passing by, everyone is going somewhere in a hurry. Each New Yorker seems to follow the Hero’s Journey: hearing a call to adventure, then following Joseph Campbell’s cycle to make it real. New York is the ultimate anti-Buddhist city, because attachment to goals reigns supreme.

Silicon Valley: a good life is … having an impact

When you meet someone at Stanford or at a Bay area meetup, the first question is often aimed at finding out whether you run a company. If not, the asker quickly loses interest, unless you can convince them that you are impressive otherwise. People (including me) are attracted to Silicon Valley because it’s a place where people discuss big ideas, and want to have an impact in the world—creating “a dent in the universe.”

Cambridge, Massachusetts: a good life is … learning together

In Cambridge, people seem to care less about what you do, and more about what you think. People—many of whom are graduate students at some of the world’s best higher education institutes—are curious to learn, and to explore topics together. I’ve learned about the way hummingbirds flap wings and how religions grow, all in a single conversation.

Amsterdam: a good life is … <many definitions possible>

It’s difficult for me to define a good life in Amsterdam. For some friends, a good life is defined by a promising career with a top-tier consulting firm or big corporate. For some friends, a good life is defined by starting creative projects. For some friends, a good life is defined by building a startup. One thing that almost all my Dutch peers value is travel—that’s why you see Dutchies everywhere around the world.

Bangalore, India: a good life is … spending time with friends

In India, more than in the United States or Europe, spending time with friends is important. When you walk the streets of Bangalore or Chennai, you see people everywhere chatting: at the chai-stand, while buying vegetables on the market, or simply standing outside. I think this culture will disappear as the country becomes more Western—Bangalore is already more “rushed” than a smaller city like Jaipur.

Have you experienced the places above differently? Which other places have distinct philosophies? Where do you live now, and does the local philosophy fit yours?


Of course, the local philosophies above are generalizations. In each city, you can find many different groups of people (i.e., tribes), each with their own philosophies. I still believe the location is important, though. In Aspen, the surrounding mountains call you to a life outdoors; in New York City, the never-ending bustle makes it natural to spend time with other people.

The most important questions in the world?

The point is to live everything. Live the questions now. Perhaps then, someday far in the future, you will gradually, without even noticing it, live your way into the answer

—Rainer Maria Rilke

Last weekend, I had breakfast with my dad. I explained that I feel particularly satisfied on days when I create something—a blog, a drawing, or a beautiful graph—and share it with someone. My dad smiled. Then, he replied, “I have a good day when something unexpected happens.”

As he said this, I realized that how you evaluate your days determines the person you become.

There are many questions we can ask ourselves daily, such as:

  • Did I create something meaningful?
  • Did I learn something new?
  • Did I make someone smile?
  • Did I get better at my craft?
  • Did I meet someone new?
  • Have I become a wiser person?
  • Did I encounter something unexpected?
  • Did I help somebody?
  • Did I do something that scared me?
  • Did I work on something that can change the world?
  • Did I invest in the people around me?
  • Did I work with amazing people?

It’s worth asking yourself: am I asking the right questions? There is no one-question-fits-all, but each question reflects certain values. Asking, “have I helped someone?” means you value compassion and kindness. Summing up what you’ve learnt at the end of the day indicates a commitment to personal growth.

Which questions do you ask yourself at the end of your day? Do they reflect what you truly care about, or are they merely a product of your environment and the past?

Thanks to Jan Overgoor for reviewing an earlier version of this post.

Why are electric grids different than the internet?

Observation #1: Information is automatically stored; electricity is not

Digital information creation commonly includes storage. When you write an email or take a picture, the information you create is automatically stored in the recording device, in the cloud, or in a combination thereof. Because creation is paired to storage, information can be consumed at another time than it was produced. You don’t need to read a book as the author writes it.

For electricity, the story is different. When a flow of electrons is created it must be transported and consumed instantaneously. Today’s gas power plants or wind turbines – electricity generation devices – do not offer the possibility to store electricity for later use. Neither do fridges or microwaves – electricity consumption devices.

N.B. The notable exception to integration of information creation and storage are spoken conversations. When we chat, our voices are not automatically stored. (This is changing, though.) Non-digital forms of information – a written letter, a painting, a Beatles’ vinyl – are stored, but are not easily replicable (see the next observation).

Observation #2: Information can be copied; electricity can not

When you send me a postcard, I can read it once, ten times or a hundred times, without the quality or quantity of the information changing. The information can be viewed an infinite number of times. When you send me a digital postcard, not only can I read it infinitely, but others can read it an infinite number of times too. Digital information is endlessly replicable – its quality and quantity doesn’t change.

A quantity of energy can only be used once, much like a kg of gold can only be used once. The physical properties change when you use the electricity. But, gold can be reused. I can not conceive of ways to derive the services from electricity without using the electrons. This is a big difference to distribution of information.


Scalability of electricity, versus scalability of information

You can visualize this difference by envisioning a ratio that equals the number of times something is used over the number of times something is created.

For electricity, the consuming versus creating ratio can be no larger than 1. Every kWh of electricity generated can only be used once (or less, if it’s wasted somewhere between generation and consumption).

For information, the consuming versus creating ratio can be much larger than 1. Every line of words typed by you can be viewed by a billion users, who read it hundreds of times.

Observation #3: Digital information can have an enormous variation in value; electricity can not

This may be the most important insight of this entire post. For a given quantity of digital information – say, 10MB of sound – the quality can range from terrible (the sound of a jetplane if your goal is to relax) to outstanding (a symphony orchestra recording for the same goal). The value for that piece of information can range from negative (I’d pay you to remove the sound) to very valuable (worth €10 per iTunes Album). Combined with digital information’s replicability (the previous observation), the large variation in value explains why software can be worth hundreds of dollars (Adobe Suite) or nothing.

In my view, electricity does not have a large distribution of value. For a given quantity of electricity, the quality is more or less equal. There can be a difference in value, depending on whether the electricity is generated close to the location of desired use or far away and whether electricity is generated according to the user’s preferences or not, but this value difference is marginal compared to the value difference of information – in the order of tens of percents.

The Future of Manufacturing – a dialogue

Tesla Factory

The text below is a transcript from a discussion with Marin Licina and Pieter Verhoeven 25-10-2013, continued by email last week. Comments have been edited for brevity and relevance. All errors are the author’s.

Titiaan: The ability to produce is becoming accessible to more and smaller groups of people. I see a future in which I have access to small-scale production technologies that manufacture food, energy or electronic devices. Will markets for produced goods continue to exist if we can make everything ourselves?

Marin: Today’s manufacturers of commodities are in jeopardy when (1) the “production recipe” is public information; (2) the raw inputs are available and (3) the production machines are affordable and accessible. Think of an electricity provider. When intangible value of a product comes in, the prediction becomes more difficult. For example, a 5 dollar quartz watch tells better time than a Rolex. From a purely functional standpoint a Rolex is a very expensive way to learn what time it is. Yet, many people want the Rolex as a status symbol: it’s expensive, hand-made by a craftsman and made of ‘precious’ metals. Status is one reason why people will not produce their own goods in a future where decentralized production is more economical than centralized production.

Pieter: Two other reasons why consumers may not produce their own goods are that consumers have a desire for a social shopping experience; and that making your own goods consumes more time than buying them from a third party.

TitiaanHow will the shift from production by few to production by many play out in industries where products need certification? Think of medicine. Field trials of medicines are conducted to get a permission to sell to the world. If everyone can make their own medicine, will certification be based on the chemical recipe of the medicine?

Pieter: Interesting question. I think some marketplaces will always be monitored by governments, and so certification will always remain a part of these marketplaces. In the example of producing your own medicine, certification will probably be based on the chemical recipe.

Marin: Perhaps what happened in the digital music industry is an interesting analogy for what we can expect with decentralized production. First, there was total anarchy: Napster, Russian download sites, anything goes. From the chaos, standards emerged: look at the iTunes store and Spotify. This is because pure anarchy didn’t yield the best results. Apple invented a better system, with more order, and like a power law, people flocked to this best system. The winner got bigger, thereby attracting even more customers, which created a virtuous cycle.

What I’m trying to say is that I expect the same for distribution of future products and recipes to make them. There will be winners, who will become the future standard. The anarchy will re-shuffle the players and create new rules. I believe order and chaos need each other.

Titiaan: I believe that when means of production shift from few to many, the information needed for production will be created rapidly. Our challenge in my view is not to open the production blueprints (digital designs), but to democratize the means of production (machines, materials). Once the means of production are democratized and people are connected, information will start to flow.

Marin: I agree with you that production is becoming increasingly decentralized. 20 years ago, only Louis Vuitton made LV bags. Now, there are a bunch of factories in China that do very good fakes. This is a problem for the likes of Vuitton and Rolex: their business model is based on having a monopoly on the design blueprint for a product. Imagine what’s going to happen when everyone can print an LV bag or Prada shoe.

Friendships as a calibrator for life

From Apple’s dictionary:

calibrate |ˈkaləˌbrāt|

• carefully assess, set, or adjust (something abstract): the regulators cannot properly calibrate the risks involved | (as adj.calibrated) : their carefully calibrated economic policies.

Earlier this week I was in Berlin. I had two wonderful conversations: one with a dear friend who I hadn’t seen for a year; the other with a woman I had never met before.

When you meet a friend you have not seen in a long time, it seems easier to talk about deep topics than with friends you see very regularly. Time together is perceived as more precious, because rarer, hence you want to use every minute to speak about stuff that matters.

There’s a second reason why long-distance friendships hold much value. Friends who see you only once every so often naturally maintain a distant perspective on your life. They don’t know about the details of every project you undertake. When these friends listen carefully and ask critical questions, such occasional conversations are a reality check: are your actions aligned with what you say your values and dreams are? These friendships serve as a calibrator for life.

How can you guarantee you have these conversations, these check-ins, to make sure you’re living a life you’re proud of?

One answer, I think, is to set time apart with friends – close-by or far-away – in which you start by discussing the very basics (your principles, your beliefs) to the very acute (what are you doing today?).

A picture is worth 1.77 million words

When I snap a picture with my iPhone 4S, the size or resolution of the image is 3264 pixels by 2448 pixels, about 8 million pixels or 8 megapixels. Stored in JPEG format (the most common compression method for digital photographs) every pixel can show one of 256 distinct colors. To allow for 256 colors you need to store 8 positions that can hold either a 0 or a 1. Two bits code for blue, three bits code for green, three bits code for red: R R R G G G B B. For example, 0 0 0 0 0 0 0 0 indicates the color black, 0 0 0 1 1 1 0 0 indicates pure green, because all green pixels take a value of 1.

The mountains I photograph with my iPhone thus require 3264 x 2448 x 8 = 64 million bits or 8 million bytes  or 8MB (1 byte = 8 bits).


Let’s take a look at text. Most languages based on the Western alphabet use between 20 to 40 letters, plus 10 numbers and some punctuation marks. 64 characters is sufficient to account for most pieces of text. For every character – for example, the letter “a”, 6 bits (2^6 = 64) should be sufficient. If an average word has 6 letters, one word requires 36 bits (6 letters x 6 bits/letter) or 4.5 bytes. One thousand words equal approximately 4500 bytes. (This is 4.5kb. Did you ever notice when saving a text file how little computer memory it uses compared to storing a picture?).

 The memory required to store a typical picture can also be used to store 1.7 million words (8 x 10^6 bytes / 4.5 bytes per word)! 

[edit: Jeroen Offerijns corrected me saying that computers use 8 bits/letter because of encoding standards. 8 bits equal 1 byte, meaning one-thousand words with on average 6 letters per word require 6,000 bytes of memory. Thus, one iPhone 4s picture equals ~1.3M words (8 x 10^6 bytes / 6 bytes per word).] 

Our brains have evolved to capture the richness of a picture practically instantly – we do not disaggregate a picture of a parrot into a matrix of 0s and 1s. If our goal is to minimize time spent on learning or amusement, eyeing at a picture is  more efficient than reading a thousand words.

For digital storage, the story is different. It is much more efficient to store millions of words (equaling 10s of MBs) than tens of high-quality images (equaling 100s of MBs); the millions of words likely provide more information too. If uploading information to our augmented brains is part of our future – words will be the clear winner over images.