How to read the genome and build a human being | Riccardo Sabatini

譯者: 易帆 余 審譯者: Jianan(Tiana) Zhao

For the next 16 minutes, I'm going to take you on a journey

接下來的16分鐘， 我要帶各位進行一段冒險之旅，

that is probably the biggest dream of humanity:

這大概是人類最大的夢想：

to understand the code of life.

了解生命的密碼。

So for me, everything started many, many years ago

對我而言，這一切的開始， 要拉回到好幾好幾年前，

when I met the first 3D printer.

當我第一次遇上3D印表機時。

The concept was fascinating.

它的概念真的很棒。

A 3D printer needs three elements:

3D印表機需要三個元素：

a bit of information, some raw material, some energy,

少量的資訊、一些原物料、 再加上點能量，

and it can produce any object that was not there before.

這樣它就可以製造出 以前從未存在過的任何東西。

I was doing physics, I was coming back home

我當時研究的是物理學， 有天回到家裡時，

and I realized that I actually always knew a 3D printer.

我突然意識到，我家裡 就有一台 3D 印表機。

And everyone does.

而且每個人家裡都有一台。

It was my mom.

那就是我媽嗎。

(Laughter)

（笑聲）

My mom takes three elements:

我媽也有三個元素：

a bit of information, which is between my father and my mom in this case,

少量的資訊：我這個例子， 指的是我媽跟我爸之間的投入，

raw elements and energy in the same media, that is food,

食物就是原物料及能量的來源，

and after several months, produces me.

然後，幾個月後，生下了我。

And I was not existent before.

而我以前也是不存的。

So apart from the shock of my mom discovering that she was a 3D printer,

所以，除了我發現我媽 就是一台3D列印機之外，

I immediately got mesmerized by that piece,

我突然間也被 這個吸引注了，

the first one, the information.

那邊的第一項，資訊。

What amount of information does it take

要有多少這樣的資訊

to build and assemble a human?

才能建構並組裝出一個人來呢？

Is it much? Is it little?

要很多嗎？還是只要一點點？

How many thumb drives can you fill?

要多少隨身碟存取這些資訊呢？

Well, I was studying physics at the beginning

我一開始是研究物理學的，

and I took this approximation of a human as a gigantic Lego piece.

我喜歡把人類比喻成 一個大型的樂高玩具，

So, imagine that the building blocks are little atoms

你可以想像，每一個 樂高積木就是一個原子，

and there is a hydrogen here, a carbon here, a nitrogen here.

氫原子在這，碳原子在這， 氮原子在這。

So in the first approximation,

按照最初的估算想法，

if I can list the number of atoms that compose a human being,

如果我可以列出 人類的原子清單的數量，

I can build it.

我就可以把它建造出來。

Now, you can run some numbers

現在，請各位算一下，

and that happens to be quite an astonishing number.

這想必是個驚人的數字。

So the number of atoms,

所以，存在這隨身碟裡面

the file that I will save in my thumb drive to assemble a little baby,

可以組合出來一個小寶寶的檔案， 裡面的原子數數量，

will actually fill an entire Titanic of thumb drives --

實際上若用樂高玩具 組裝起一個人類，

multiplied 2,000 times.

它的大小足足有 2000台鐵達尼號這麼大。

This is the miracle of life.

這就是生命的奇蹟啊！

Every time you see from now on a pregnant lady,

從現在起，你每次 看到懷孕的婦女，

she's assembling the biggest amount of information

她就是那個正在組裝

that you will ever encounter.

你這輩子所遇到的最大量資訊。

Forget big data, forget anything you heard of.

忘了大數據吧！ 忘了你曾聽過的。

This is the biggest amount of information that exists.

這就是現存的 最大數據資料。

(Applause)

（笑聲）

但...好在大自然比一位 年輕的物理學家還聰明，

But nature, fortunately, is much smarter than a young physicist,

and in four billion years, managed to pack this information

這40億年來，大自然中 負責管理包裹這個資訊的

in a small crystal we call DNA.

小晶體--我們稱之為DNA。

We met it for the first time in 1950 when Rosalind Franklin,

我們在1950年第一次認識了它，

an amazing scientist, a woman,

當時有一位了不起的女科學家 --羅莎琳．富蘭克林--

took a picture of it.

給 DNA 拍了張照。

But it took us more than 40 years to finally poke inside a human cell,

但我們花了40年的時間， 最後才戳進人類細胞裡

take out this crystal,

取出這個晶體，

unroll it, and read it for the first time.

才首次把它伸展開來閱讀。

The code comes out to be a fairly simple alphabet,

而密碼也就是大家所孰知的

four letters: A, T, C and G.

四個字母：A、T、C、G。

And to build a human, you need three billion of them.

而建造一個人類， 你需要30億個字母。

Three billion.

30億。

How many are three billion?

30億有多少？

It doesn't really make any sense as a number, right?

我們對這個數字 真的很沒有概念，對吧？

So I was thinking how I could explain myself better

所以，我在想，這麼大的數字

我要怎麼解釋 才讓人比較容易了解。

about how big and enormous this code is.

But there is -- I mean, I'm going to have some help,

但，我的意思是... 我最好找個人來幫忙，

and the best person to help me introduce the code

而能幫我介紹基因密碼 的最佳人選，

is actually the first man to sequence it, Dr. Craig Venter.

想當然就是第一個定序的人， 克萊格．凡特博士。

So welcome onstage, Dr. Craig Venter.

所以，讓我們歡迎 克萊格．凡特博士上台。

(Applause)

（掌聲）

Not the man in the flesh,

當然不是活生生的人，

but for the first time in history,

但這是史上第一次

this is the genome of a specific human,

特定人類的基因組被

printed page-by-page, letter-by-letter:

一頁接著一頁，一個字 接著一個字地列印出來：

262,000 pages of information,

262,000頁的資料，

450 kilograms, shipped from the United States to Canada

450公斤、從美國運到加拿大，

thanks to Bruno Bowden, Lulu.com, a start-up, did everything.

感謝新創公司Lulu.com的布魯諾．鮑登， 他們幫我做的這一切。

It was an amazing feat.

這是個很棒的饗宴。

But this is the visual perception of what is the code of life.

但這只是對生命密碼 的視覺感受。

And now, for the first time, I can do something fun.

現在，為了慶祝第一次， 我要做件有趣的事。

I can actually poke inside it and read.

我真的可以從裡面 挑一段來讀一讀。

So let me take an interesting book ... like this one.

所以，讓我來找一本有趣的.... 書兒，比如這本。

I have an annotation; it's a fairly big book.

我做了個註記；這書太厚了。

So just to let you see what is the code of life.

讓各位看一下甚麼是生命密碼。

Thousands and thousands and thousands

數以百萬、千萬、

and millions of letters.

億個字母。

And they apparently make sense.

它們當然都有意義。

Let's get to a specific part.

讓我來找一段特別的

Let me read it to you:

讀給各位聽：

(Laughter)

（笑聲）

"AAG, AAT, ATA."

"AAG, AAT, ATA."

To you it sounds like mute letters,

你們可能覺得像是在聽天書，

but this sequence gives the color of the eyes to Craig.

但這段序列，決定了 克萊格的眼睛顏色。

I'll show you another part of the book.

我再展示另一段給各位看。

This is actually a little more complicated.

這段實際上稍微複雜些。

Chromosome 14, book 132:

14 號染色體，第132 號書：

(Laughter)

（笑聲）

As you might expect.

如你所望！

(Laughter)

（笑聲）

"ATT, CTT, GATT."

"ATT, CTT, GATT."

This human is lucky,

這個人很幸運，

because if you miss just two letters in this position --

因為如果你在這個位置 剛好漏掉兩個字母--

two letters of our three billion --

30億個字母，只漏掉兩個--

he will be condemned to a terrible disease:

你就等同於被宣判 得了一個恐佈的疾病：

cystic fibrosis.

囊性纖維化。

We have no cure for it, we don't know how to solve it,

目前我們沒有治療的方式， 我們不知道如何解決，

and it's just two letters of difference from what we are.

僅僅就這兩個字母上 的差異而已。

A wonderful book, a mighty book,

這本偉大的書，

a mighty book that helped me understand

這本偉大的書，

可以幫助我了解，也能讓各位 看到一些嘆為觀止的事情。

and show you something quite remarkable.

Every one of you -- what makes me, me and you, you --

在場的每一個人， 成就你我不同的地方

is just about five million of these,

就這五百萬個 字母的差異，

half a book.

半本書。

For the rest,

剩下的，

we are all absolutely identical.

我們絕對都長一樣。

Five hundred pages is the miracle of life that you are.

就是這 500 頁的字母， 行塑了你是甚麼樣的人，

The rest, we all share it.

剩下的，我們都一樣。

So think about that again when we think that we are different.

所以，當我們在討論彼此差異的時候， 讓我們再反思一下，

This is the amount that we share.

其實我們共同的地方 真的有這麼多。

So now that I have your attention,

所以，我問一下各位，

the next question is:

接下來的問題：

How do I read it?

我要怎麼讀它？

How do I make sense out of it?

我要怎麼搞懂它？

Well, for however good you can be at assembling Swedish furniture,

其實，無論你多麼會 看說明書組裝瑞典的家具，

this instruction manual is nothing you can crack in your life.

這本安裝手冊也沒辦法 教你如何破解你的人生。

(Laughter)

（笑聲）

And so, in 2014, two famous TEDsters,

2014年，兩位出名的 TED 演講者，

Peter Diamandis and Craig Venter himself,

彼得．戴曼迪斯和 克雷格．文特爾本人，

decided to assemble a new company.

他們決定創立一家新公司。

Human Longevity was born,

《人類長壽公司》誕生了，

with one mission:

並賦予一個使命：

trying everything we can try

竭盡所能的，

and learning everything we can learn from these books,

從這些書上，嘗試每樣東西， 學習每樣東西，

with one target --

就為了一個目標——

making real the dream of personalized medicine,

讓個人化醫療的美夢可以成真，

understanding what things should be done to have better health

了解需要做哪些事 才能更健康，

and what are the secrets in these books.

以及了解這些書 裡面的秘密。

An amazing team, 40 data scientists and many, many more people,

一個令人驚豔的團隊，40 個數據科學家， 還有其他很多、很多的人，

a pleasure to work with.

一起為團隊努力。

The concept is actually very simple.

這概念其實很簡單。

We're going to use a technology called machine learning.

我們將要使用一種叫 「機械自主學習」的概念。

On one side, we have genomes -- thousands of them.

一方面，我們有 成千上萬的基因組——

On the other side, we collected the biggest database of human beings:

另一方面，我們收集了 人類最大的資料庫：

phenotypes, 3D scan, NMR -- everything you can think of.

生物特性、3D掃描、核磁共振—— 你能想到的每樣東西。

Inside there, on these two opposite sides,

這兩方面的資料， 被自主翻譯出來後

there is the secret of translation.

就可以解開很多的祕密。

And in the middle, we build a machine.

在這兩個中間， 我們建立了一台機器。

We build a machine and we train a machine --

我建立它，訓練它——

well, not exactly one machine, many, many machines --

當然，並不只一台機器啦！ 是很多很多台機器——

to try to understand and translate the genome in a phenotype.

嘗試去了解並翻譯 基因組的生物特徵表象。

What are those letters, and what do they do?

這些字母代表甚麼？ 它們有甚麼作用？

It's an approach that can be used for everything,

這個方法可以運用在每件事上，

but using it in genomics is particularly complicated.

但用在基因學上， 它就特別複雜。

Little by little we grew and we wanted to build different challenges.

在一點一滴的慢慢累積後， 我們想建立不一樣的挑戰。

We started from the beginning, from common traits.

我們從共同的特徵開始。

Common traits are comfortable because they are common,

談共同特徵比較輕鬆， 因為它們都很普遍。

everyone has them.

每個人都有。

So we started to ask our questions:

我們從這個問題開始問：

Can we predict height?

我們可以預測身高嗎？

Can we read the books and predict your height?

我們可以光看書 就可以知道你的身高嗎？

Well, we actually can,

沒錯，我們真的可以，

with five centimeters of precision.

預測的誤差在五公分內。

BMI is fairly connected to your lifestyle,

身體質量指數與 你的生活形式有關，

but we still can, we get in the ballpark, eight kilograms of precision.

但我們仍然可以，相當精準地 將預測誤差控制在 8 公斤以內。

Can we predict eye color?

那我們可以預測眼睛顏色嗎？

Yeah, we can.

是的，我們可以。

Eighty percent accuracy.

精準度高達80%。

Can we predict skin color?

我們可以預測皮膚顏色嗎？

Yeah we can, 80 percent accuracy.

是的，可以，80%的準確率。

Can we predict age?

年齡呢？

We can, because apparently, the code changes during your life.

可以，因為隨著年紀， 你的基因碼也會更著改變。

It gets shorter, you lose pieces, it gets insertions.

它會變短、消失或被插入。

We read the signals, and we make a model.

我們可以讀到那個訊號， 並把它模擬出來。

Now, an interesting challenge:

現在，有一項有趣的挑戰：

Can we predict a human face?

我們可以預測一個人的臉嗎？

It's a little complicated,

這有點複雜，

because a human face is scattered among millions of these letters.

因為人臉上散播了 上百萬個這種字母。

And a human face is not a very well-defined object.

而人臉不太容易預測。

So, we had to build an entire tier of it

所以，我們必須建立一個 完整的堆疊系統，

to learn and teach a machine what a face is,

去學習並教會機器 人臉是甚麼，

and embed and compress it.

然後把它嵌進去並壓縮。

And if you're comfortable with machine learning,

如果你很懂機器自主學習，

you understand what the challenge is here.

你會懂得這邊的挑戰是甚麼。

Now, after 15 years -- 15 years after we read the first sequence --

15年後--整整15年後-- 我們讀取到第一個序列--

this October, we started to see some signals.

今年10月，我們開始看到一些訊號。

And it was a very emotional moment.

真的是令人感動的時刻。

What you see here is a subject coming in our lab.

你現在看到的是一個 進來我們實驗室的實驗對象。

This is a face for us.

這是一個我們人類的臉。

So we take the real face of a subject, we reduce the complexity,

所以我們拿一個真實的臉當作實驗對象， 我們減少了複雜度，

because not everything is in your face --

因為不是每樣東西都會在 你的臉上原貌呈現出來--

lots of features and defects and asymmetries come from your life.

有很多的特徵、缺陷及不對稱 來自於你後天的生活方式。

We symmetrize the face, and we run our algorithm.

我們把臉對稱好後， 拿去跑我們的演算法。

The results that I show you right now,

我現在展示給各位看的結果，

this is the prediction we have from the blood.

是由血液演算出來的預測結果。

(Applause)

（掌聲）

Wait a second.

稍等一下。

In these seconds, your eyes are watching, left and right, left and right,

在這短短的幾秒鐘，你的眼睛會 左看看、右看看做比較，

and your brain wants those pictures to be identical.

而你的大腦會希望 這些照片是一致的。

So I ask you to do another exercise, to be honest.

所以，我要求各位做另一項活動， 這次要誠實。

Please search for the differences,

請找出他們不一樣的地方，

which are many.

有很多喔。

The biggest amount of signal comes from gender,

最多的訊號來自性別，

then there is age, BMI, the ethnicity component of a human.

然後是年齡、身體質量指數、 人類種族族群。

And scaling up over that signal is much more complicated.

把這些訊號擴大是相當複雜的。

But what you see here, even in the differences,

但即使你現在看到有點不同，

lets you understand that we are in the right ballpark,

還是要讓各位知道， 我們預測還算不錯，

that we are getting closer.

已經很接近了。

And it's already giving you some emotions.

這已經讓你有點激動了。

This is another subject that comes in place,

這裡有另外一個例子，

and this is a prediction.

這是預測的結果。

A little smaller face, we didn't get the complete cranial structure,

有點小的臉，我們雖然沒有 跑完整個頭蓋骨結構，

but still, it's in the ballpark.

但，還是很精準。

This is a subject that comes in our lab,

這是另一個實驗對象，

and this is the prediction.

這是預測結果。

So these people have never been seen in the training of the machine.

這些人從未在我們 訓練的機器裡面出現過。

These are the so-called "held-out" set.

也就是說這些從 外面隨機取樣的。

But these are people that you will probably never believe.

但也許各位不相信。

We're publishing everything in a scientific publication,

我們已經在科學期刊上 發表這一切了，

you can read it.

你可以找到。

But since we are onstage, Chris challenged me.

但自從知道我們要上台後， 克里斯就挑戰我說，

I probably exposed myself and tried to predict

我也許可以自己上陣

someone that you might recognize.

並嘗試預測你們可能認識的人。

So, in this vial of blood -- and believe me, you have no idea

所以，在這一瓶血液裡面-- 相信我，你們絕對不知道

what we had to do to have this blood now, here --

我們去哪裡搞來這一瓶血的，

in this vial of blood is the amount of biological information

這瓶血就擁有 全部的生物資訊，

that we need to do a full genome sequence.

夠我們跑完全部的基因組定序。

We just need this amount.

我們只需要這麼多。

We ran this sequence, and I'm going to do it with you.

我們已經把它拿去定序， 下次再做給大家看。

And we start to layer up all the understanding we have.

然後開始堆疊出 所有我們知道的東西，

In the vial of blood, we predicted he's a male.

從這瓶血液裡， 我們預測出他是位男士。

And the subject is a male.

而實驗對象是男士。

We predict that he's a meter and 76 cm.

我們預測他身高176公分。

The subject is a meter and 77 cm.

實際上他身高177公分。

So, we predicted that he's 76; the subject is 82.

我們預測他的體重是76公斤； 實際上是82公斤。

We predict his age, 38.

我們預測他的年齡是38歲。

The subject is 35.

實際上是35歲。

We predict his eye color.

我們預測眼睛的顏色是這樣。

Too dark.

太暗了。

We predict his skin color.

我們預測他的皮膚顏色。

We are almost there.

幾乎很接近了。

That's his face.

這是他的臉。

Now, the reveal moment:

現在，真相要大白的時刻了：

the subject is this person.

他長這樣。

(Laughter)

（笑聲）

And I did it intentionally.

我故意這樣做的。

I am a very particular and peculiar ethnicity.

我是一個非常特別的奇特種族。

Southern European, Italians -- they never fit in models.

南歐洲人、義大利人—— 他們從來不會跟我們的預測相符。

And it's particular -- that ethnicity is a complex corner case for our model.

這個種族在我們的模式下， 就是一個很複雜的特殊案例。

But there is another point.

但有另外一個重點。

So, one of the things that we use a lot to recognize people

我們用很多工具 來辨認人的特徵，

will never be written in the genome.

但絕對不會把這些特徵 寫到基因組裡面。

It's our free will, it's how I look.

因為這是我們的自由意志， 我就是長這樣。

Not my haircut in this case, but my beard cut.

在這個案例中，重點不是我的髮型， 而是我的鬍鬚。

So I'm going to show you, I'm going to, in this case, transfer it --

所以，我要秀給各位看， 我會把它轉變一下--

and this is nothing more than Photoshop, no modeling --

就僅是用Photoshop上個鬍子，

the beard on the subject.

沒有調整其他的。

And immediately, we get much, much better in the feeling.

突然間，感覺就比較像了。

So, why do we do this?

所以，我們為什麼要做這個？

We certainly don't do it for predicting height

我們絕對不是為了預測高度

or taking a beautiful picture out of your blood.

或拍一張你血液的美麗照片。

We do it because the same technology and the same approach,

我們這樣做的原因是， 這些科技、方法、

the machine learning of this code,

機器自主學習程式，

is helping us to understand how we work,

可以幫助我們了解 我們要如何進行工作、

how your body works,

你的身體是如何運作、

how your body ages,

你的身體如何老化、

how disease generates in your body,

你身上的疾病是如何造成的、

how your cancer grows and develops,

你的癌症是如何成長和擴散的、

how drugs work

藥物如何運作、

and if they work on your body.

以及這些藥物在你身上是否有作用。

This is a huge challenge.

這是一個很大的挑戰。

This is a challenge that we share

這是我們全世界的 研究人員共同的挑戰。

with thousands of other researchers around the world.

It's called personalized medicine.

它叫做個人化醫療。

It's the ability to move from a statistical approach

這種醫療能力是從 傳統的統計方法，

where you're a dot in the ocean,

讓你大海撈針亂吃藥，

to a personalized approach,

轉成個人客製化的方法，

where we read all these books

都是從閱讀這些書裡面，

and we get an understanding of exactly how you are.

讓我們了解真正的你。

But it is a particularly complicated challenge,

但這是充滿了複雜的挑戰，

because of all these books, as of today,

因為到目前為止，這些書，

we just know probably two percent:

我們僅大概了解2%：

four books of more than 175.

四本書又175頁。

And this is not the topic of my talk,

但這不是我演講的主題，

because we will learn more.

因為我們還有很多要學。

There are the best minds in the world on this topic.

全世界最聰明的智慧 就在這個主題裡面。

The prediction will get better,

預測會越來越改善，

the model will get more precise.

模式會越來越精準。

And the more we learn,

我們學得越多，

the more we will be confronted with decisions

我們克服從未面對過 的決策的能力就越強，

that we never had to face before

about life,

有關於生命、

about death,

死亡、

about parenting.

養育的決策。

So, we are touching the very inner detail on how life works.

所以，我們正接觸到 生命如何運作的內部細節。

And it's a revolution that cannot be confined

而且這個革命不能只侷限在

in the domain of science or technology.

主流科學或技術上。

This must be a global conversation.

我們需要一個全球性的對話。

We must start to think of the future we're building as a humanity.

我們必須開始思考， 我們要建構的人類未來。

We need to interact with creatives, with artists, with philosophers,

我們需要與創意人才、 藝術家、哲學家

with politicians.

政治家相互配合。

Everyone is involved,

每個人都要參與其中，

because it's the future of our species.

因為這是我們人類的未來。

Without fear, but with the understanding

不需要害怕，但需要包容

that the decisions that we make in the next year

明年我們所做的決定，

will change the course of history forever.

將永遠地改變歷史。

Thank you.

謝謝各位！

(Applause)

（掌聲）